RX 11 Manual

Table of Contents
1. Introduction ................................................................................................................. 1
New in RX 11 ............................................................................................................. 1

2. RX Overview ................................................................................................................ 2
Navigating the manual ................................................................................................. 2
Feature comparisons .................................................................................................. 2

3. Working with Files ........................................................................................................ 6

Opening Files ............................................................................................................. 6
Supported File Formats ............................................................................................... 6
Supported Channel Configurations ................................................................................ 7
Creating New Files ...................................................................................................... 7
Managing File Tabs ..................................................................................................... 7
Saving Files ............................................................................................................... 8
Saving RX Documents ................................................................................................. 8
Export Options ........................................................................................................... 9
File Info ................................................................................................................... 13
Closing Files ............................................................................................................ 13

4. Recording in the RX 11 Audio Editor .............................................................................. 16

Overview ................................................................................................................. 16
Troubleshooting ....................................................................................................... 16

5. Transport Controls & Displays ...................................................................................... 17

Overview ................................................................................................................. 17
Transport ................................................................................................................ 17
Time and Frequency Readouts ................................................................................... 18
Transport Clock ........................................................................................................ 18
Time Format Display ................................................................................................. 18

6. Spectrogram/Waveform Display ................................................................................... 19

Overview ................................................................................................................. 19
Spectrogram Settings ................................................................................................ 20
Rulers ..................................................................................................................... 23
Waveform Displays ................................................................................................... 24
Waveform Overview .................................................................................................. 24

7. Text Navigation .......................................................................................................... 26

Overview ................................................................................................................. 26
Workflow ................................................................................................................. 26
Search .................................................................................................................... 27
Multiple Speaker Detection ......................................................................................... 28
Export Transcript ...................................................................................................... 29

8. Interactive Tools ......................................................................................................... 30

Overview ................................................................................................................. 30
Zoom Tools ............................................................................................................. 30
Navigation Tools ....................................................................................................... 31
Channel Selectors ..................................................................................................... 31
Channel Order (Multichannel) ..................................................................................... 32
Instant Process ........................................................................................................ 32
Selection tools ......................................................................................................... 33
View Clip Gain .......................................................................................................... 34
 Selection Modifiers ................................................................................................... 35

9. Undo History .............................................................................................................. 37

Overview ................................................................................................................. 37
Export History .......................................................................................................... 37
Restore Selection ...................................................................................................... 37
Expandable History List ............................................................................................. 38

10. Application Menus .................................................................................................... 39

File ......................................................................................................................... 39
Edit ........................................................................................................................ 39
View ....................................................................................................................... 40
Modules .................................................................................................................. 42
Transport ................................................................................................................ 43
Window ................................................................................................................... 44
Help ....................................................................................................................... 45

11. Preferences ............................................................................................................. 46

Audio ...................................................................................................................... 46
Display .................................................................................................................... 47
Keyboard ................................................................................................................. 48
Misc ....................................................................................................................... 49
License and Updates ................................................................................................. 50
Plug-ins ................................................................................................................... 51
Assistant ................................................................................................................. 52

12. Composite View ....................................................................................................... 53

Overview ................................................................................................................. 53
Workflow ................................................................................................................. 53
Important Notes ....................................................................................................... 53

13. Batch Processor ....................................................................................................... 55

Overview ................................................................................................................. 55
Batch Processor Presets ............................................................................................ 55
Input ....................................................................................................................... 55
Module Chain ........................................................................................................... 56
Output .................................................................................................................... 56

14. Repair Assistant ....................................................................................................... 58

Overview ................................................................................................................. 58
Repair Assistant Module ............................................................................................ 58
Repair Assistant Plug-in ............................................................................................. 59
Voice Mode ............................................................................................................. 59
Music Mode ............................................................................................................. 61
Repair Assistant Preferences ...................................................................................... 62

15. Common Module Controls ......................................................................................... 63

Overview ................................................................................................................. 63
Presets ................................................................................................................... 63
Module Footer Controls ............................................................................................. 64
Preview ................................................................................................................... 64
Compare ................................................................................................................. 65

16. Module List .............................................................................................................. 67

Overview ................................................................................................................. 67
Saving module list filters ............................................................................................ 68
17. Module Chain ........................................................................................................... 69
Overview ................................................................................................................. 69
Controls .................................................................................................................. 69

18. Ambience Match ...................................................................................................... 71

Overview ................................................................................................................. 71
Controls .................................................................................................................. 71
Ambience Type ........................................................................................................ 72
Workflow ................................................................................................................. 72

19. Breath Control .......................................................................................................... 74

Overview ................................................................................................................. 74
Controls .................................................................................................................. 74

20. Center Extract .......................................................................................................... 76

Overview ................................................................................................................. 76
Controls .................................................................................................................. 76
Use Cases ............................................................................................................... 77

21. De-bleed .................................................................................................................. 78

Overview ................................................................................................................. 78
Controls .................................................................................................................. 79
Workflow ................................................................................................................. 80

22. De-click ................................................................................................................... 81

Overview ................................................................................................................. 81
Controls .................................................................................................................. 81
Instant Process Tool ................................................................................................. 82

23. De-clip ..................................................................................................................... 83

Overview ................................................................................................................. 83
Controls .................................................................................................................. 83
More Information ...................................................................................................... 85

24. De-crackle ............................................................................................................... 86

Overview ................................................................................................................. 86
Controls .................................................................................................................. 86

25. De-ess ..................................................................................................................... 87

Overview ................................................................................................................. 87
Controls .................................................................................................................. 87

26. De-hum ................................................................................................................... 89

Overview ................................................................................................................. 89
Mode Selection ........................................................................................................ 89
Controls in Dynamic Mode ......................................................................................... 90
Controls in Static Mode ............................................................................................. 91
More Information ...................................................................................................... 93

27. De-plosive ................................................................................................................ 95

Overview ................................................................................................................. 95
Controls .................................................................................................................. 95

28. De-reverb ................................................................................................................. 97

Overview ................................................................................................................. 97
Controls .................................................................................................................. 97
More Information ...................................................................................................... 99
29. De-rustle ................................................................................................................ 102
Overview ............................................................................................................... 102
Controls ................................................................................................................ 102

30. De-wind ................................................................................................................. 104

Overview ............................................................................................................... 104
Controls ................................................................................................................ 104
Tips ...................................................................................................................... 105
Alternative Modules ................................................................................................ 105

31. Deconstruct ........................................................................................................... 106

Overview ............................................................................................................... 106
Controls ................................................................................................................ 106
More Information .................................................................................................... 107

32. Dialogue Contour .................................................................................................... 108

Overview ............................................................................................................... 108
Displays ................................................................................................................ 108
Contour Curve Editing .............................................................................................. 110
Controls ................................................................................................................ 111
Alternative Modules ................................................................................................ 112

33. Dialogue Isolate ...................................................................................................... 113

Overview ............................................................................................................... 113
Controls (STD and ADV) ........................................................................................... 113
Controls (ADV only) ................................................................................................ 114
Alternative Modules ................................................................................................ 115

34. Guitar De-noise ....................................................................................................... 116

Overview ............................................................................................................... 116
Controls ................................................................................................................ 116
Processing Tips ...................................................................................................... 119
Alternative Modules ................................................................................................ 119

35. Interpolate ............................................................................................................. 120

Overview ............................................................................................................... 120
Controls ................................................................................................................ 120

36. Loudness Optimize ................................................................................................. 122

Overview ............................................................................................................... 122
Metering and Readouts ........................................................................................... 122
Controls ................................................................................................................ 123
Learn .................................................................................................................... 123

37. Mouth De-click ....................................................................................................... 125

Overview ............................................................................................................... 125
Controls ................................................................................................................ 125
More Information .................................................................................................... 125

38. Music Rebalance .................................................................................................... 126

Overview ............................................................................................................... 126
Controls ................................................................................................................ 126
Stem Split .............................................................................................................. 127
Getting started with Music Rebalance in Logic ............................................................ 127

39. Spectral De-noise ................................................................................................... 130

Overview ............................................................................................................... 130
 Controls ................................................................................................................ 130
Advanced Settings .................................................................................................. 133
More Information .................................................................................................... 137

40. Spectral Repair ....................................................................................................... 138

Overview ............................................................................................................... 138
Attenuate .............................................................................................................. 138
Replace ................................................................................................................. 139
Pattern .................................................................................................................. 140
Partials and Noise ................................................................................................... 140
Surrounding Region Display ...................................................................................... 141
Workflow ............................................................................................................... 141
More information .................................................................................................... 142

41. Spectral Recovery ................................................................................................... 144

Overview ............................................................................................................... 144
Controls ................................................................................................................ 144

42. Streaming Preview .................................................................................................. 146

Overview ............................................................................................................... 146
Controls ................................................................................................................ 146
Workflow and Presets ............................................................................................. 147

43. Voice De-noise ....................................................................................................... 148

Overview ............................................................................................................... 148
Controls ................................................................................................................ 148
Voice De-noise Plug-in ............................................................................................. 149

44. Wow & Flutter ......................................................................................................... 150

Overview ............................................................................................................... 150
Controls ................................................................................................................ 150

45. Azimuth ................................................................................................................. 153

Overview ............................................................................................................... 153
Controls ................................................................................................................ 153
More Information .................................................................................................... 154

46. Dither .................................................................................................................... 155

Overview ............................................................................................................... 155
Controls ................................................................................................................ 155

47. EQ ........................................................................................................................ 157

Overview ............................................................................................................... 157
Controls ................................................................................................................ 157
More Information .................................................................................................... 158

48. EQ Match ............................................................................................................... 159

Overview ............................................................................................................... 159
Controls ................................................................................................................ 159
Workflow ............................................................................................................... 159

49. Fade ...................................................................................................................... 161

Overview ............................................................................................................... 161
Controls ................................................................................................................ 161
Instant Process Tool ............................................................................................... 161

50. Gain ...................................................................................................................... 162

 Overview ............................................................................................................... 162
Controls ................................................................................................................ 162
Instant Process Tool ............................................................................................... 162

51. Leveler .................................................................................................................. 163

Overview ............................................................................................................... 163
Controls ................................................................................................................ 164

52. Loudness Control .................................................................................................... 166

Overview ............................................................................................................... 166
Controls ................................................................................................................ 166
Workflow ............................................................................................................... 167
Loudness Standards ............................................................................................... 167

53. Mixing ................................................................................................................... 168

Overview ............................................................................................................... 168
Controls ................................................................................................................ 168

54. Normalize .............................................................................................................. 169

Overview ............................................................................................................... 169
Target Peak Level ................................................................................................... 169

55. Phase .................................................................................................................... 170

Overview ............................................................................................................... 170
Controls ................................................................................................................ 170
More Information .................................................................................................... 170

56. Plug-in Hosting ....................................................................................................... 172

Overview ............................................................................................................... 172
Selection Based Processing ..................................................................................... 172
Plug-in Presets ....................................................................................................... 172
Assign Presets to Keyboard Shortcuts ....................................................................... 173

57. Resample .............................................................................................................. 174

Overview ............................................................................................................... 174
Controls ................................................................................................................ 174

58. Signal Generator ..................................................................................................... 177

Overview ............................................................................................................... 177
Silence .................................................................................................................. 177
Tones ................................................................................................................... 177
Noise .................................................................................................................... 178

59. Time & Pitch .......................................................................................................... 179

Overview ............................................................................................................... 179
iZotope Radius ....................................................................................................... 179
Controls ................................................................................................................ 179
More Information .................................................................................................... 182

60. Variable Pitch ......................................................................................................... 183

Overview ............................................................................................................... 183
Displays ................................................................................................................ 183
Contour Curve Editing .............................................................................................. 185
Controls ................................................................................................................ 187
Compare Settings ................................................................................................... 188
Alternative Modules ................................................................................................ 188
61. Variable Time ......................................................................................................... 189
Overview ............................................................................................................... 189
Displays ................................................................................................................ 189
Contour Curve Editing .............................................................................................. 191
Controls ................................................................................................................ 192
Compare Settings ................................................................................................... 193
Alternative Modules ................................................................................................ 193

62. Find Similar ............................................................................................................ 194

Overview ............................................................................................................... 194
Controls ................................................................................................................ 194

63. Markers & Regions .................................................................................................. 195

Overview ............................................................................................................... 195
Controls ................................................................................................................ 195
More Information .................................................................................................... 195

64. Spectrum Analyzer .................................................................................................. 196

Overview ............................................................................................................... 196
Controls ................................................................................................................ 196

65. Waveform Statistics ................................................................................................ 197

Overview ............................................................................................................... 197
Controls ................................................................................................................ 197
More Information .................................................................................................... 198

66. RX Plug-ins ............................................................................................................ 199

Plug-in Formats ...................................................................................................... 199
Learning in Audiosuite ............................................................................................. 199
Presets ................................................................................................................. 200
History .................................................................................................................. 201
I/O Meters ............................................................................................................. 201
Options ................................................................................................................. 201
Latency ................................................................................................................. 202

67. RX Monitor ............................................................................................................. 203

Overview ............................................................................................................... 203
Workflow ............................................................................................................... 203

68. RX Connect ............................................................................................................ 204

Overview ............................................................................................................... 204
Controls ................................................................................................................ 204
DAW/NLE specific instructions ................................................................................. 204
Using RX as an External Audio Editor ......................................................................... 207

69. RX Spectral Editor ................................................................................................... 209

Overview ............................................................................................................... 209
Getting Started with Spectral Editor (ARA) .................................................................. 209
Troubleshooting ..................................................................................................... 212
Controls ................................................................................................................ 212
Processing Modes .................................................................................................. 213
Working with Comp Takes in Logic Pro ....................................................................... 217
Keyboard Shortcuts ................................................................................................ 217
ARA Requirements and Limitations ........................................................................... 217

70. Keyboard Shortcuts ................................................................................................ 219

71. Identifying Audio Problems ...................................................................................... 228
RX Processing Step Flowchart .................................................................................. 228
Using the Spectrogram Display ................................................................................. 230
Hum ..................................................................................................................... 230
Buzz ..................................................................................................................... 230
Hiss and other Broadband Noise ............................................................................... 231
Clicks, Pops, & Short Impulse Noises ......................................................................... 231
Clipping ................................................................................................................. 231
Intermittent Noises ................................................................................................. 232
Gaps and Drop Outs ................................................................................................ 233

72. Authorization ......................................................................................................... 234

Trial Mode ............................................................................................................. 235
Demo Mode ........................................................................................................... 235
Authorization Methods ............................................................................................ 236
More Authorization Help .......................................................................................... 236

73. License Information ................................................................................................ 237

Abseil ................................................................................................................... 237
Anti-Grain Geometry ................................................................................................ 237
ARA ...................................................................................................................... 237
ARM_NEON_2_x86_SSE ........................................................................................... 238
base64 .................................................................................................................. 238
Better Enums ......................................................................................................... 239
Boost .................................................................................................................... 239
Bravura ................................................................................................................. 240
C++ Rest SDK ........................................................................................................ 241
Caffe .................................................................................................................... 243
cerberus ................................................................................................................ 244
composable_kernel ................................................................................................. 244
coremltools ........................................................................................................... 245
CpuInfo ................................................................................................................. 246
Crashpad .............................................................................................................. 246
curl ....................................................................................................................... 247
dcleblanc/SafeInt ................................................................................................... 247
Distributed Machine Learning Common Codebase ....................................................... 247
dlfcn-win32 ............................................................................................................ 248
DLPack ................................................................................................................. 248
DLPack: Open In Memory Tensor Structure ................................................................. 248
Eigen .................................................................................................................... 248
emsdk .................................................................................................................. 249
FarmHash ............................................................................................................. 249
FLAC .................................................................................................................... 249
FlashAttention ........................................................................................................ 250
FlatBuffers ............................................................................................................ 250
FreeBSD: getopt.c file .............................................................................................. 251
FreeType ............................................................................................................... 251
FXDev ................................................................................................................... 251
G3log: Asynchronous logger with Dynamic Sinks ......................................................... 252
gemmlowp ............................................................................................................ 252
GLEW ................................................................................................................... 252
google/glog ........................................................................................................... 254
google/nsync ......................................................................................................... 255
google/re2 ............................................................................................................. 255
google/sentencepiece ............................................................................................. 255
google/XNNPACK ................................................................................................... 256
gsl ........................................................................................................................ 256
HalidelR ................................................................................................................ 257
HowardHinnant/date ............................................................................................... 257
huggingface/transformers ....................................................................................... 258
IcoMoon ............................................................................................................... 258
Intel Math Kernel Library (Intel MKL) .......................................................................... 258
Intel neural-compressor ........................................................................................... 259
intel/dnnl ............................................................................................................... 260
Intel® Integrated Performance Primitives (Intel® IPP) .................................................. 260
JDAI-CV/DNNLibrary ............................................................................................... 261
JsonCpp ............................................................................................................... 261
LAME .................................................................................................................... 261
libb64 ................................................................................................................... 262
libprotobuf-mutator ................................................................................................. 262
LibXML2 ............................................................................................................... 262
madler/zlib ............................................................................................................ 263
magic_enum .......................................................................................................... 263
Microsoft Cognitive Toolkit (CNTK) ............................................................................ 264
microsoft/mimalloc ................................................................................................ 264
microsoft/wil ......................................................................................................... 265
mpi4py ................................................................................................................. 265
MurmurHash3 ........................................................................................................ 266
Netlib numeralgo na10 Aberth’s method ..................................................................... 266
nlohmann/json ....................................................................................................... 267
nvidia/cutlass ........................................................................................................ 267
OGG / Vorbis .......................................................................................................... 267
onnx ..................................................................................................................... 268
onnx/onnx-tensorrt ................................................................................................. 268
onnxruntime .......................................................................................................... 269
Open MPI .............................................................................................................. 269
Opus .................................................................................................................... 271
PIL ....................................................................................................................... 272
portmidi ................................................................................................................ 272
posix pthread library ................................................................................................ 273
protocolbuffers/protobuf ......................................................................................... 273
pthreadpool ........................................................................................................... 274
Pytorch / Caffe2 ..................................................................................................... 274
pytorch/cpuinfo ...................................................................................................... 275
react-native ........................................................................................................... 276
readerwriterqueue .................................................................................................. 276
Resemblyzer .......................................................................................................... 277
Roboto font family .................................................................................................. 277
ruy ....................................................................................................................... 277
Skia ...................................................................................................................... 278
SQLite ................................................................................................................... 278
TagLib .................................................................................................................. 278
Tencent/rapidjson .................................................................................................. 279
Tensorflow ............................................................................................................ 280
TinyXML ................................................................................................................ 280
Tipue Search .......................................................................................................... 281
vectorize ............................................................................................................... 281
WebView2 ............................................................................................................. 282
xsimd ................................................................................................................... 282
Yoga ..................................................................................................................... 283
ZeroMQ ................................................................................................................. 283
zlib ....................................................................................................................... 284
 INTRODUCTION 1

1. Introduction

Welcome to RX 11!
RX 11 features new machine learning algorithms for dialogue isolation and dialogue de-reverb that
work in real-time. We’ve added new tools to preview audio for streaming and to optimize loudness.
Additionally, there are major improvements to Repair Assistant, as well as higher-quality source
separation. We've also expanded the list of compatible hosts for the Spectral Editor ARA plugin.

New in RX 11
• Dialogue Isolate: Combines dialogue de-noising and de-reverbing in a single, realtime plugin
and in an updated module. The Advanced plugin and module also include multi-band
processing and a new offline high-quality mode.
• Streaming Preview: Hear music the way that the rest of the world will by applying the same
loudness normalization and lossy data encoding that streaming platforms will apply.
• Loudness Optimize: Optimize perceived loudness and minimize measured LUFS of your track
prior to posting it on streaming platforms with level normalization.
• Music Rebalance: A new neural net provides a largely improved quality of musical source
separation as well as per-stem Sensitivity adjustment.
• Repair Assistant: The addition of machine learning technology provides better processing and
faster learning of your audio.
• Dialogue Contour: Change the character and expressivity of a voice with new Formant and
Variation controls, and see those changes reflected get in the new pitch curve.
• Mid/Side Channel Mode: Visualize and edit audio in Mid/Side configuration in the main RX
window.
• Signal Generator: New Pulses tone shape, new Modulate pasting mode.
• Preferences: True peak calculation mode allows selection of the oversampling algorithm for
maximal compliance of true peak level measurements.
 RX OVERVIEW 2

2. RX Overview
iZotope’s award-winning RX Audio Editor is the industry standard for audio repair, restoration, and
enhancement. It offers a comprehensive suite of tools focused on alleviating common to complex
audio issues. Post production professionals, audio engineers, and video editors alike use RX to
transform problematic recordings into production-ready audio.

Navigating the manual

This help guide is shared by RX 11 Standard and RX 11 Advanced. The following tag is used
throughout the manual to differentiate Advanced only features: ADV

Feature comparisons
The following tables outline the differences between Standard and Advanced. The rightmost
column in each of the tables indicates features that are new or improved in RX 11.

RX 11 Audio Editor Feature Comparison

Features Standard Advanced New/Improved

Batch Processor X X
Clip Gain X X
Composite View X X
Expandable History list X X
Find Similar X X
First Time User Tour X X Improved
Instant Process X X
Markers & Regions X X
Mid/Side Mode X X New
Module Chain X X
Module List View Filters X X
MP3 Export X X
Multichannel Processing X
Plug-in Hosting X X
Recording & Monitoring X X
Repair Assistant X X Improved
Restore Selection X X
Spectrum Analyzer X X
Spectral Editing Tools X X
Text Navigation X X
Time/Frequency Feathering X X
Waveform Statistics X X
 RX OVERVIEW 3

RX 11 Module Comparison

Module Name Standard Advanced New/Improved

Ambience Match X
Azimuth X
Breath Control X X
Center Extract X
De-bleed X X
De-click X X
De-clip X X
De-crackle X X
De-ess X X
De-hum X X
De-plosive X X
De-reverb X X
De-rustle X
De-wind X
Deconstruct X
Dialogue Isolate with De-reverb X X New
Dialogue Contour X Improved
Dither X X
EQ X X
EQ Match X
Fade X X
Gain X X
Guitar De-noise X X
Interpolate X X
Leveler X
Loudness Control X X
Loudness Optimize X X New
Mixing X X
Mouth De-click X X
Music Rebalance X X Improved
Normalize X X
Phase X X
Repair Assistant X X Improved
Resample X X
Signal Generator X X
Spectral De-noise X X
Spectral Recovery X
Spectral Repair X X
 RX OVERVIEW 4

Module Name Standard Advanced New/Improved

Streaming Preview X X New
Time & Pitch X X
Variable Pitch X X
Variable Time X X
Voice De-noise X X
Wow & Flutter X

RX 11 Plug-in Comparison
The RX 11 installer will give you the option to install plug-ins in the following formats:
• AU
• AAX
• AAX Audiosuite

RX 11 Ambience Match and De-rustle plug-ins are AAX Audiosuite for Pro Tools
only.

• VST3
• AU ARA
• VST3 ARA
• AAX ARA

Spectral Editor ARA is available in Logic Pro (AU), StudioOne (VST3), and Pro
Tools (AAX, via an upcoming update from Avid). Music Rebalance ARA is only
compatible with Logic Pro (AU).

All plug-in formats are 64 bit only. VST2 is no longer supported.

RX Plug-ins Elements Standard Advanced New/Improved

Ambience Match X
Breath Control X X
Connect X X
De-click X X X
De-clip X X X
De-crackle X X
De-ess X X
De-hum X X X
De-plosive X X
De-reverb X X X
De-rustle X
Dialogue Isolate X X Improved
Guitar De-noise X X
 RX OVERVIEW 5

RX Plug-ins Elements Standard Advanced New/Improved

Monitor X X
Mouth De-click X X
Music Rebalance X X Improved
Repair Assistant X X X Improved
Spectral De-noise X X
Spectral Editor X X Improved
Voice De-noise X X X
 WORKING WITH FILES 6

3. Working with Files

Opening Files
RX supports opening up to 32 audio files at a time. Files can be opened in the RX Audio Editor
using the following methods:
1. Navigate to the File menu, select Open… and choose the files from the system dialog that
appears. Alternatively, the following keyboard shortcuts can be used to launch the Open…
system dialog: Command+O (Mac) or ctrl+O (Windows).
2. Drag and drop files into the main editor window to open them in a new file tab.
3. Drag a file from Finder/Windows Explorer onto the RX Audio Editor icon in the Dock/Desktop.
4. Click on the Open file button that appears in the RX Audio Editor window when no files are
loaded.
5. Double-click on the RX logo in the middle of the RX Audio Editor interface when no files are
loaded.

Supported File Formats

A number of different file formats can be opened and edited in the RX Audio Editor. The next three
sections outline the supported file formats and channel count configurations that can be opened in
the RX Audio Editor. For information about file formats when saving or exporting files in RX, see the
Saving Files and Export Options sections below.

Supported Audio File Formats

The following audio file formats can be opened in the RX Audio Editor: AAC, AAX (Audible
Audiobook Format), AIFF/AIF, BWF, CAF, FLAC, M4A, MP3, OGG, SD2, WAV, WMA

Supported Video File Formats

The RX Audio Editor supports loading video file formats, but does not support video playback.
When a video file is opened in RX, only the audio data from that file will be imported.
The following video file formats can be opened in the RX Audio Editor: AVI, M4V, MOV (RX requires
that Quicktime is installed in order to open Quicktime file formats, e.g. .mov files.), MPEG, MPV,
WMV.

FILE FORMAT DEPENDENCIES

Some file formats may have dependencies based on your operating system that may
prevent you from importing them into the RX Audio Editor. For example, Windows
native formats (like WMA and WMV) may not open on Mac and QuickTime formats
(like AAC, MOV, and M4V) may require installing QuickTime on Windows and running
the RX Audio Editor in 32-bit mode.
 WORKING WITH FILES 7

Supported Channel Configurations

The RX 11 Standard Audio Editor and RX 11 Advanced Audio Editor support opening mono and
stereo audio files.

TIP: OPTION FOR OPENING SPLIT STEREO FILES IN ONE TAB

Mono audio files with (.L and .R) or (.1 and .2) extensions can be opened as either
mono files (2 mono tabs) or split stereo (1 stereo file tab). See Misc for more
information. Note that this option is only applicable to split stereo files and does not
apply to split surround files.

Multichannel File Support

The RX 11 Advanced Audio Editor supports opening audio files with up 10 channels per file tab.
Multichannel audio device settings can be configured in the “Audio” tab of the Preferences menu.
The channel selector labels can be configured by selecting an option in the Channel Order menu.
To access the Channel Order menu, right-click on the Time ruler and navigate to the “Channel
Order” sub-menu. The options available in the Channel Order menu depend on the number of
channels in the active file tab.

Creating New Files

To create a new file in RX:
1. Open the File menu
2. Select “New…”
3. You will be prompted for the name, sample rate and channel count of the new file you are
creating.

TIP: CREATE A NEW FILE FROM THE CONTENTS OF THE CLIPBOARD

If you have existing audio data in your clipboard (for example, if you have copied a
selection from an existing file in RX), you can create a new file based on that audio
data.
Open the “File” menu, choose “New from Clipboard” or use the keyboard shortcut:
Command+Shift+N (Mac) or Ctrl+Shift+N (Windows). The new file will match the
sample rate and channel count of the audio data present on your clipboard.

Managing File Tabs

RX supports having up to 32 files open at once. You can navigate between tabs by clicking on a tab
or using the following keyboard shortcuts:
• Select tab to the right of current tab: Control+Tab (Mac) or Alt+Tab (Windows)
• Select tab to the left of current tab: Control+Shift+Tab (Mac) or Alt+Shift+Tab (Windows)
If you have multiple files open, an arrow button ( ) will appear to the right of the last visible tab.
You can access file tabs that are not currently visible by clicking on the arrow button and selecting
a tab from the menu.
 WORKING WITH FILES 8

▶ Close any tab by clicking it with the center mouse button.

Saving Files
There are a number of ways to save a file in the RX Audio Editor. The Save Operations include:

Name Description Default Mac Default Windows

Shortcut Shortcut
Save For uncompressed file Command+S Ctrl+S
formats (.wav or .aiff):
Overwrites the original file on
disk
For compressed file Command+S Ctrl+S
formats: Opens the Export
File dialog
Save As… For uncompressed file Command+Shift+S Ctrl+Shift+S
formats (.wav or .aiff): Save
a copy of your file using the
same file format
For compressed file Command+Shift+S Ctrl+Shift+S
formats: Opens the Export
File dialog
Save RX Saves file as .rxdoc file
Document extension (more information
below)
Save RX Saves copy of your .rxdoc file
Document
As…

AUTOSAVE
The RX Audio Editor will automatically save backups of your editing session by
default. When the RX application is launched, it will open your most recent editing
session. The option to turn it off is located under the Preferences > Misc tab as
“Resume last editing session when app starts.”

Saving RX Documents
You can save a file using the RX Document file format (.rxdoc) to archive your edits. An RX
Document includes your original file, all the edits you’ve made to it, and your most recent selection
and view state. RX Documents can only be opened in the RX Audio Editor. If you need to save
your file so it can be opened somewhere else (like a DAW or media player), you need to export it in
another format (like WAV or AIFF).
To save an RX Document, select File > Save RX Document… and select where you would like to
store the file. Keep in mind that the size of the RX Document file can be very large, especially if your
list of edits include multiple processes on the whole file.
 WORKING WITH FILES 9

Export Options
There are a number of different export options in the RX Audio Editor:
• Export File
• Export Selection
• Export Regions to Files
• Export Screenshot
• Export History as XML

Export File

1. Select File… > Export

2. Select the file format you want to Export to and adjust the associated settings as desired
(available settings explained in the table below)
3. Click “OK”
4. In the system window, name your file and choose where you would like to save it to
5. Click “Save” to export your file. Checking the Reopen file in RX checkbox will open your
exported file in the RX 11 Audio Editor after the export completes successfully.

Export Format Options

The following file formats are available when exporting files from the RX 11 Audio Editor:
• WAV - Uncompressed
• AIFF - Uncompressed
• FLAC - Compressed: Lossless Compression
• OGG - Compressed: Lossy Compression
• MP3 - Compressed: Lossy Compression

Uncompressed File Formats

The RX 11 Audio Editor allows you to export files to the following uncompressed file formats: WAV
and AIFF.

WAV
The following options are available when exporting files to this file format:
• Bit Depth: Determines the bit depth of the exported file. Choices include: 16 bit, 24 bit, 32 bit
(float), 32 bit (int).
 WORKING WITH FILES 10

• Dither: Determines the Dither Type to be applied to the exported file. Choices include: None,
White Noise (TPDF), Noise shaping (MBIT+)
• BWF: When selected, the file will be exported with extended information that is included in the
file header of a Broadcast Wave File.
• Preserve Non-Audio Data: When selected, the exported file will retain the metadata of the
original file.

AIFF
The following options are available when exporting files to this file format:
• Bit Depth: Determines the bit depth of the exported file. Choices include: 16 bit, 24 bit, 32 bit
(float), and 32 bit (int).
• Dither: Determines the Dither Type to be applied to the exported file. Choices include: None,
White Noise (TPDF), and Noise shaping (MBIT+).
• Preserve Non-Audio Data: When selected, the exported file will retain the metadata of the
original file.

Compressed File Formats

The RX 11 Audio Editor allows you to export files to the following compressed file formats: FLAC
(Lossless), OGG (Lossy), and MP3 (Lossy).

FLAC (Lossless)
The FLAC file format offers lossless compression. The following options are available when
exporting files to this file format:
• Bit Depth: Determines the bit depth of the exported file. Choices include: 8 bit, 16 bit, and 24 bit
• Dither: Determines the Dither Type that will be applied to the exported file. Choices include:
None, White Noise (TPDF), and Noise shaping (MBIT+).
• Compression Level: Adjusts the compression strength of the FLAC encoder. Stronger
compression requires more CPU time during file encoding but results in a slightly smaller file.
FLAC compression setting does not result in any quality change to the signal since FLAC is a
lossless format.

OGG (Lossy)
The Ogg Vorbis file format offers lossy compression. The following options are available when
exporting files to this file format:
• Quality: Adjusts the bitrate of the Ogg Vorbis compression algorithm. Higher bitrate values
result in higher audio quality, but also increase the file size
• Prevent Clipping: See the Prevent Clipping section below for more information.

MP3 (Lossy)
The mp3 file format offers lossy compression. The following options are available when
exporting files to this file format:
• Bit rate: Adjusts the bit rate of the MP3 compression algorithm. Higher bit rates result in higher
quality audio but will increase the file size.
• Bit rate control: Determines how (or if) bit rate varies over time. Choices include: Constant bit
rate (CBR), Average bit rate (ABR), and Variable bit rate (VBR).
• Prevent Clipping: See the Prevent Clipping section below for more information.

Prevent Clipping
Predicts and prevents codec clipping when exporting audio in lossy formats (MP3 and OGG) by
checking for decoded levels and adjusting levels of the original signal.
 WORKING WITH FILES 11

PREVENT CLIPPING PROCESSING TIME

Prevent Clipping may run significantly slower than regular encoding, since it
computes the correct level adjustment depending on the amount of clipping occurring
in the file. Files with little to no codec clipping usually will encode quickly, whereas
heavily clipping files may take longer.

There are two types of file level adjustments that can be applied:
• Normalize: attenuates overall level of the file to ensure that the encoded/decoded file does not
exceed 0 dBTP.
• Limiter: attenuates parts of the file that could become clipped to retain the level of non-
clipping sections, while overall true peak levels are limited to 0 dBTP.

CHOOSING NORMALIZE OR LIMITER

The Limiter will leave larger sections of the file unchanged in level and will
only attenuate sections that would experience clipping. However, like any dynamic
processing, this may create pumping. The Normalize mode can completely avoid
pumping at the expense of slightly reducing the overall level of the file.

Export Selection
This option will allow you to export only the audio that is contained within your current selection, as
opposed to the entire audio file.
1. Select File > Export Selection, and the Export File dialogue box appears.
2. Follow the additional aforementioned steps.

Export Regions to Files

This option allows you to export each region in the currently selected file to an individual audio file.
1. Open the File menu and choose Export Regions to Files…
2. Choose the export file format in the Export Files dialog and click “OK”
 WORKING WITH FILES 12

3. In the Export Regions to Files system dialog:

a. Choose the destination for the exported files. By default, the destination is set to the
location of the source file.
b. Enter a suffix to append to all exported region filenames. When the default suffix of (none)
is used, a suffix will not be added to the exported file names.

NOTE
The names of the exported files will match the names of the regions with the
optional suffix appended to the filename. If any regions share the same name,
a number will be appended sequentially to the exported filenames to avoid
duplication.

4. Click Save to export.

Export Screenshot
This option allows you to export your current Spectrogram/Waveform display as a PNG image file.
This can be very helpful for archiving any restoration process or for forensic documentation.

When clicking on Export Screenshot from the File menu, your current Spectrogram/Waveform view
will be used for adjusting your screenshot size and position.

The Spectrogram/Waveform transparency balance must be set before selecting File >
Export Screenshot as this cannot be changed in this window.

To define the size of your screenshot, simply click and drag in order to enlarge or shrink
the screenshot window. The dimensions of your resulting screenshot will update automatically,
however these can also be entered manually by clicking once in either Width or Height.

The max resolution attainable for your screenshot will be limited by the individual
computer’s screen resolution.
 WORKING WITH FILES 13

When you are finished changing the dimensions of your screenshot, click on the Save button to
name and save your .PNG screenshot to your chosen directory.

To save screenshots faster (at the expense of having a larger file on disk), disable
Maximum image compression.

Export History as XML

Export the Undo history list of your current file tab to an .xml document.

File Info
The File Info window can be opened by navigating to Window menu > File Info. There are
two sections in the File Info window: General Info and More Info. The More Info section lists
information dependent on the file type. The following information is available in the General Info
and More Info sections of the File Info window:
• General Info
• Name: The current filename
• Duration: Length of the file
• Sampling rate: The original sampling rate of the file
• Bit depth: The original bit depth of the file
• Channels: Mono or stereo
• Size on disk: Size of the file in bytes
• More Info
• Timecode
• Created by
• Originator reference
• Date created
• Time created
• BWF version
• Coding history
• Track Title
• Artist
• Album
• Date
• Track Number
• Comment
• Genre

Closing Files
The following sections describe different methods availble for closing file tabs in the RX Audio
Editor.
 WORKING WITH FILES 14

Close One File Tab

Single file tabs can be closed using the following methods:
• Single-click on the ‘x’ button in the file tab display
• Select the “Close file” option in the File menu
• Right-click on any file tab and select “Close” from the context menu

• Keyboard shortcuts: Command+W (Mac) or ctrl+W (Windows)

Close Other File Tabs

To keep one file tab open and close all other file tabs: Right-click on the file tab that should remain
open and select “Close others” from the context menu.

Close All File Tabs

All file tabs can be closed using the following methods:
• Select the “Close all files” option in the File menu
• Right-click on any file tab and select “Close all” from the context menu
• Keyboard shortcuts: Command+Shift+W (Mac) or ctrl+Shift+W (Windows)

Closing File Tabs With Unsaved Changes

If a file has been edited or processed in the RX Audio Editor and the changes have not been saved,
a small dot will appear in the corner of the file tab to indicate that there are unsaved changes.
When closing file tabs that have unsaved changes, a prompt will be displayed before the file
is closed. The prompt will include options to save, revert changes or cancel before closing the
modified file.

The following options are available in the prompt:

• Yes: The modified file will be saved as an RX Document file (.rxdoc), a system window will
appear to select the save location for the file before closing the tab.
• No: Unsaved changes will be discarded and the file tab will be closed.
• Cancel: The prompt will be dismissed and the file tab will remain open.

Closing The Application With File Tabs Open

The RX Audio Editor application will open all file tabs present when it was last closed if the “Reopen
previous audio files when app starts” option in the Preferences > Misc tab is enabled.
 WORKING WITH FILES 15

If this option is enabled and the application is closed when files with unsaved changes are present,
a prompt will not be displayed. Any unsaved changes will be stored in the RX session data folder
and will load the next time the application is opened.
If this option is disabled, a prompt will appear to save or discard changes when closing the
application if any file tab has unsaved changes. A separate prompt will appear for each file tab with
unsaved changes. If any of the prompt dialogs are canceled, the application will remain open.
 RECORDING IN THE RX 11 AUDIO EDITOR 16

4. Recording in the RX 11 Audio Editor

Overview
RX supports recording up to two channels at a time.
To record in the RX 11 Audio Editor:
1. Create a new file.
2. Press the Record button once to arm recording. The Record button will flash red when RX is
armed to record. The meters to the right of the transport controls will update based on your
input signal when recording is armed.
3. Before recording, you should ensure that your input levels are not clipping and allow for
adequate headroom. Alternatively, you can enable input monitoring to set input levels without
engaging record arm.
4. After adjusting your input levels, you can start recording by clicking the Record button again.
When RX is recording, the Record button will display as solid red.
5. You can stop recording by clicking the Record button again.
6. After you have stopped recording you can edit and apply processing to the file.

RX 11 SESSION DATA FOLDER

After recording, your recorded audio data is stored in the RX 11 Session Data folder.
You can set the location of the RX Session Data folder in Misc. If you use the
recording functionality in the RX Audio Editor often, it is recommended that the RX
Session Data folder be located on a drive with a sufficient amount of free space.

Troubleshooting
If you are having trouble recording in RX, try the following steps:
1. Enable Input Monitoring and look for activity on RX’s level meters.
2. Close other audio applications, DAWs and NLEs open on your computer to make sure no other
program is usurping the sound card.
3. Open Preferences > Audio and make sure the correct device is listed in Input Device. Also
check in the Channel Routing dialog to make sure the correct inputs are selected.
4. Check your input source. Make sure the hardware connections between whatever you’re
recording from and the inputs on your audio interface are correct.
 TRANSPORT CONTROLS & DISPLAYS 17

5. Transport Controls & Displays

Overview
The following sections include information about the different transport controls, readouts, and
displays you will encounter in the RX Audio Editor application.

Transport

Name Description
INPUT MONITOR When enabled, allows you to monitor input signal to set levels
prior to recording. Input source is configured in the RX Audio
Editor Preferences
RECORD Begins recording into a new file. One click puts recording into an
armed state for safely setting input levels. The next click begins
recording. A third click stops recording. If you already have a
file open, hitting Record will prompt you to create a new file for
recording.
REWIND [Enter/ Brings you back to the start of the file.
Return]
PLAY [Spacebar] Starts and stops playback. Starts or stops recording if Record is
armed.
PLAY SELECTION When you’ve made a selection of a time range, frequency range,
ONLY or both, this button auditions just the selection (useful for
isolating intermittent noises, etc.)
LOOP Ctrl+L Enable this switch to loop the selected audio.
(Windows) Cmd+L
(Mac)
PLAYHEAD Toggles the behavior of the playhead on stop. If this is enabled,
FOLLOWS the playhead will return to the anchor sample (the position
PLAYBACK Ctrl+R before playback began).
(Windows) Cmd+R
(Mac)
PLAYHEAD To place the playhead, single click anywhere in the
Spectrogram/Waveform display. To playback audio while
positioning the playhead, click and drag the playhead icon or
hold Ctrl (Windows) or Command (Mac) while clicking and
dragging in the Spectrogram/Waveform display.
 TRANSPORT CONTROLS & DISPLAYS 18

Time and Frequency Readouts

• The time display (on the left in the image above) shows the start, end and length time values
of the current selection and the current view. The time format used in this display is selected in
the time format menu, explained in the section below.
• The frequency display (on the right in the image above) shows the low frequency boundary,
high frequency boundary and frequency range values for the current selection and the current
view range.
• Clicking on any of these fields allows you to manually enter values.

Transport Clock
Indicates the current position of the playhead. Depending on time format display setting, this value
is shown in hours/minutes/seconds, time code, or samples.

Time Format Display

You can change the time format display used in the time ruler and time readouts. It can be
accessed by right-clicking the time ruler or by clicking the arrow button to the left of the currently
selected time format label.

Format Name & Description

Samples: The sample counter, starting from 0

Time (h:m:s): Time in hours, minutes, seconds, and milliseconds, starting from
0
Timecode (n fps): The time code in hours, minutes, seconds, and frames,
starting from 0
Source Time (h:m:s): Time in hours, minutes, seconds, and milliseconds,
starting from the clip’s timecode origin
Source Timecode (n fps): The time code in hours, minutes, seconds, and
frames, starting from the clip’s timecode origin. The value of n (the frame rate of
the time code) is determined by the Time Scale Frame Rate setting in the Misc
tab of the Preferences window
 SPECTROGRAM/WAVEFORM DISPLAY 19

6. Spectrogram/Waveform Display
Overview
The RX Audio Editor features a rich visual environment for editing and repairing audio.
The central focus of the interface is the Spectrogram/Waveform display. It combines an
advanced Spectrogram with a waveform transparency overlay to provide frequency and amplitude
information in one highly configurable window.
Using the spectrogram to identify audio problems
See the Identifying Audio Problems chapter for tips on using the spectrogram to spot common
audio issues.

Anatomy of the Spectrogram Display

The spectrogram allows you to visualize both frequency and amplitude information of an audio
recording in one display.

Frequency
The Spectrogram shows frequency information across the vertical axis. Lowest frequency content
is displayed at the bottom, highest frequency content is displayed at the top.

This image shows the spectrogram of a sine sweep over pink noise. The sine sweep starts at 20 Hz
(bottom of the display) and sweeps to 20 kHz (top of the display) over 4 minutes.

Amplitude and Color

The amplitude of frequency content is indicated by variations in color in the Spectrogram. The
color map ruler (to the right of the frequency ruler) shows the color being used to represent a given
amplitude value.
 SPECTROGRAM/WAVEFORM DISPLAY 20

In this example, Louder events (speech) are indicated by brighter colors (yellow/bright orange) and
quieter events (breaks in speech and noise floor) are indicated by darker colors (dark orange, blue,
black)

Spectrogram Settings
The RX Spectrogram is highly configurable, you can adjust the default configuration, load a preset
or save your own preset in the Spectrogram Settings window.
The Spectrogram Settings window can be opened using the following methods:
• From the “View” menu of the RX Audio Editor.
• By right-clicking on the spectrogram display and selecting Spectrogram Settings from the
context menu.
• Using a keyboard shortcut: command+shift+, (on Mac) or ctrl+shift+, (on Windows).

Spectrogram Type
RX offers different methods for displaying time and frequency information in the Spectrogram.
RX’s advanced Spectrogram modes allow you to see sharper time (horizontal) and frequency
(vertical) resolution simultaneously. There is always a trade-off of display quality versus processing
time, so keep in mind that some modes will take longer to draw on the screen than others.
 SPECTROGRAM/WAVEFORM DISPLAY 21

TYPE DESCRIPTION
REGULAR STFT Most common spectrogram type (can be found in other editors) It
has a fixed uniform time-frequency resolution. This is the simplest and
fastest drawing mode in RX.

AUTO- Automatically adjusts FFT size (i.e. time and frequency resolution of a
ADJUSTABLE STFT Spectrogram) according to the zoom level. For example, if you zoom in
horizontally (time) you’ll see that percussive sounds and transients will
be more clearly defined. When you zoom in vertically (frequency), you’ll
see individual musical notes and frequency events will appear more
clearly defined.

MULTI- Calculates the Spectrogram with better frequency resolution at low

RESOLUTION frequencies and better time resolution at high frequencies. This
mimics psychoacoustic properties of our perception, allowing the
Spectrogram display to show you the most important information
clearly.

ADAPTIVELY Automatically varies the time and frequency resolution of the

SPARSE Spectrogram to achieve the best Spectrogram sharpness in every area
of the time-frequency plane. This often lets you see the most details
for a thorough analysis, but it’s the slowest mode to calculate.

FFT Size
The greater the FFT size, the greater the frequency resolution, i.e. notes and tonal events will
be clearer at larger sizes. However, choosing a larger number here will make time events less
sharply defined because of the way this type of processing is done. Choosing Auto-adjustable or
Multi-resolution modes allows you to get a good combination of frequency and time resolution
without having to change this setting as you work.
Fast Fourier Transform (FFT)
A procedure for the calculation of a signal frequency spectrum. The greater the FFT size, the
greater the frequency resolution, i.e., notes and tonal events will be clearer at larger sizes.

Enable Reassignment
Enables a special technique for Spectrogram calculation that allows very precise pitch tracking
for any harmonic components of the signal. When used together with Frequency Overlap/
Time Overlap controls, this option can provide virtually unlimited time and frequency resolution
simultaneously for signals consisting of tones.
 SPECTROGRAM/WAVEFORM DISPLAY 22

Window
Selects between the different weighting functions (or windows) that are used for the FFT analysis.
Window functions control the amount of signal leakage between frequency bins of the FFT. “Weak”
windows, such as Rectangular, allow a lot of leakage, which may blur your Spectrogram vertically.
“Strong” windows, such as Kaiser or cos3, eliminate leakage at the expense of a slight loss of
frequency resolution.

Frequency Scale
Frequency scale adjustments can help you see useful information more easily. Different scales
have different characteristics for displaying the vertical (frequency) information in the Spectrogram
display.
• LINEAR: Displays frequencies spread out in a uniform way. This is most useful when you want
to analyze higher frequencies.
• LOGARITHMIC: This scale puts more attention on lower frequencies.
• MEL: the Mel scale (derived from the word Melody) is a frequency scale based on how humans
perceive sound. This selection is one of the more intuitive choices because it corresponds to
how we hear differences in pitch.
• BARK: The Bark scale is also based on how we perceive sound, and corresponds to a series of
critical bands.

Frequency Overlap
Controls the amount of oversampling on the frequency scale of Spectrogram. When used together
with the Reassignment option, it will increase the resolution of the Spectrogram vertically (by
frequency).

Time Overlap
This controls the time oversampling of the Spectrogram. In most cases, overlap of 4x or 8x is a
good setting to start with. However, using higher overlap together with the Reassignment option
will increase the time resolution of a Spectrogram, letting you see transient events clearly.

Color Map
The Spectrogram display allows you to choose between several different color schemes. There
is no right or wrong color setting to use and we recommend you try them all to determine your
preference. Sometimes certain color modes will make different types of noise stand out more
clearly. Experiment!
 SPECTROGRAM/WAVEFORM DISPLAY 23

High-Quality Rendering
Accurate max-bilinear interpolation of the Spectrogram (recommended). Turning this control
off makes Spectrogram rendering slightly faster, but you’ll lose some detail and clarity in the
Spectrogram image.

Reduce Quality Above

RX’s Spectrogram uses very accurate rendering, letting you see audio problems, such as clicks,
even at low zoom levels. However, performing such rendering for long files can be somewhat
slow. When the length of the visible Spectrogram is above the specified number of seconds, the
Spectrogram calculation is changed to a fast and less accurate preview mode. When you zoom in,
the Spectrogram calculation becomes accurate again.

Cache Size (MB)

Limits the amount of memory used by the Spectrogram.

Rulers
On the right side of the Spectrogram/Waveform display are the Amplitude ruler for the Waveform,
Frequency ruler for the Spectrogram, and Color Map ruler for the Spectrogram.

Amplitude Rulers
You can right-click on the spectral Amplitude ruler to reveal a selection of amplitude scales:
• dB: Shows Waveform levels in decibels, relative to digital full scale (it is the most common type
of scale used for spectrum analyzers).
• NORMALIZED: Shows Waveform levels relative to the full scale level of 1.
• 16 BIT: Shows Waveform levels as quantization steps of a 16-bit audio format (−32768 to
+32767).
• PERCENT: Shows Waveform levels as percentage from full scale.

Color Map Ruler

This ruler shows what color represents what amplitude in the Spectrogram. The range of this
display is the dynamic range of the RX Spectrogram. You can click and drag the map to change the
range and use the scroll wheel to make the range larger or smaller. This is useful for seeing very
quiet noises without using gain to change the level of your audio.

Frequency Rulers
Right-clicking on the frequency ruler will display the frequency scale options:
• LINEAR: Linear scale means that Hertz are linearly spaced on a screen.
 SPECTROGRAM/WAVEFORM DISPLAY 24

• MEL (default) & BARK: Mel and Bark are frequency scales commonly found in
psychoacoustics, and reflect how our ears detect pitch. They are approximately linear below
500 Hz and approximately logarithmic above 500 Hz.
• MEL scale reflects our perception of pitch: equal subjective pitch increments produce equal
increments in screen coordinates.
• BARK scale reflects our subjective loudness perception and energy integration. It is similar
to Mel scale, but puts more emphasis on low frequencies.
• LOG: in this mode, different octaves occupy equal screen space. The screen coordinates are
proportional to the logarithm of Hertz down to 100 Hz.
• EXTENDED LOG: this extends the logarithmic scale down to 10 Hz, so that it puts even more
attention on lower frequencies.
• PIANO ROLL OVERLAY: A representation of how specific frequency ranges correlate to the
western musical scale can be displayed by right-clicking on the Frequency ruler and selecting
Show Piano Roll. If you would like to hide the frequency indicators so they don’t obscure this
piano roll, you can disable Show Frequencies and Ticks (which is enabled by default).

Waveform Displays
The Spectrogram Display features the Waveform Transparency Balance slider that lets you
superimpose a Waveform display over the Spectrogram:

The slider lets you see both frequency and overall amplitude at the same time. This can be
invaluable for quickly identifying clipping, clicks and pops, and other events.

Waveform Overview
An overview of the entire audio file’s Waveform is displayed above the main Spectrogram/
Waveform display in order to provide a handy reference point when zooming and making audio
selections in RX.

The Waveform overview will always display the entire audio file, and will also display any selections
made in the main display. When zooming in on your audio, the currently visible audio region will
also be highlighted in the Waveform overview. Click and drag on the highlighted region in order to
scroll your main audio display left or right, and click and drag on the edges of the highlighted region
in order to make the zoom tighter or wider. To zoom out fully, simply double click on the highlighted
visible region.
 SPECTROGRAM/WAVEFORM DISPLAY 25

NOTE
Hover over the waveform overview and use your mousewheel to scale the amplitude
of the waveform display to provide a clearer overview. This will not affect the
amplitude scaling in the main Spectrogram/Waveform display.
 TEXT NAVIGATION 26

7. Text Navigation

Overview
Text Navigation converts speech into a transcription that is displayed above the spectrogram
and shown in sync with the corresponding audio. Transcribed text is searchable and provides
reference points for what is contained in the file. This eliminates the need to audition files in order
to manually place markers.

TEXT NAVIGATION NOTE

Text Navigation is designed to be an editing navigation tool and not a transcription
service. It is optimized for American English. Accuracy may vary if there is noticeable
background noise or a speaker has a non-American accent.

Workflow
To get started, drag or import audio into RX and click the Speech Recognition Word Lane button at
the bottom left of the spectrogram.

Audio transcription begins immediately and runs in the background at approximately 7 - 9X real
time. The transcription populates the tabs in the Word Lane above the spectrogram, indicating the
transcription is in process.

The audio material has to fulfill the following requirements:

• Audio must be dialogue or speech – the transcription of musical lyrics is not currently
supported.
• Audio files must be at least 10 seconds long. The speech recognition button is disabled for files
shorter than 10 seconds.
 TEXT NAVIGATION 27

• Only American English is supported at this time

Clicking the iZotope logo opens a small window that shows the progress of the transcription. At
the bottom of this window are buttons for pausing or canceling the transcription.

Once the transcription is finished, you can zoom in or out of the audio and the transcription
expands or contracts accordingly.
• Mac: Command+= to zoom in, Command+- to zoom out, or swipe up or down with two fingers
on a trackpad
• Windows: Ctrl+= to zoom in, Ctrl+- to zoom out
Click on a word tab to select the corresponding audio in the spectrogram. Drag the handles on
either side of a tab to select surrounding words, a phrase, or a sentence.
When zoomed in close enough so there is a single word per tab, double-clicking on the word
selects it and makes it editable. Type in a correction for a misspelled word or alter individual words
to fit your editing needs.
• When editing, typing multiple words in the tab does not split the tab up
• Typing nothing in the tab will not delete the tab
• Right-click the Word Lane to Rescan. (Rescanning will overwrite any corrections)

Search
Text Navigation includes a fuzzy search to find words and variants of words, such as misspellings
in the transcript. You can also search for individual letters.
Search is ideal for finding a replacement word, navigating to a specific section, or locating an
alternate take.
Searching three characters or less works like auto-complete, where the search looks for words that
begin with those letters. Searching anything else works like a fuzzy search and will try to return
results that are similar to the search query.
Click on the Text Navigation Pane button to bring up the search panel.
 TEXT NAVIGATION 28

NOTE
The Word Lane button will also need to be enabled in order to see the transcription
above the spectrogram.

Type a word in the search box and hit return/enter. If the word is found, every instance of it will be
listed in the order in which it appears in the audio. Variants of the word are also listed. If the word
isn’t found, variants are listed only if they are identified in the audio.
Clicking on a word in the list moves the playhead to that word in the transcription and highlights
the corresponding audio.
If you have a word that needs replacing, search for instances of that word, pick the best one, then
copy it and paste it over the original.
Search can be used to target processing to a selected word. Dragging the handles on either side of
the word targets processing to the selected audio.

NOTE
Search works with edited words, but typing more than a single word or phrase in a tab
may negatively impact search results.

Multiple Speaker Detection

Built into Text Navigation is functionality that automatically detects when there is more than one
speaker on a track and color codes the sections of speech associated with each speaker.
Multiple speaker detection runs after the text transcription pass has finished. Up to 8 speakers can
be detected.
Each speaker receives a unique identifying color, which is indicated in the speaker pane and in the
corresponding tabs of the transcription.
 TEXT NAVIGATION 29

To select all instances of a speaker, click on the speaker name in the speaker pane. This makes it
easy to target specific processing by speaker.
Double-click a speaker’s name to edit or change it to fit the particular needs of a project.

Export Transcript
To access the Transcript Export menu, click the menu button at the top of the Text Navigation
Pane or right-click in the Word Lane.

• Copy transcript to clipboard: Copy the transcribed text and paste it into a word processing
application.
• Export transcript to file: Export the transcribed text as a .txt file.
• Rescan speech to text: Transcribe your file again.
 INTERACTIVE TOOLS 30

8. Interactive Tools

Overview
Below the Spectrogram display is a toolbar that includes several options for working with the
spectrogram/waveform display. The toolbar is split into three main categories, Navigation, Instant
Process, and Selection.

Zoom Tools
You can zoom in horizontally and vertically on both the waveform and spectrogram views.

Zoom Selection Tools

The following zoom selection tools are available in the RX Audio Editor:

• Zoom in: zooms in on the time ruler.

• Zoom out: zooms out on the time ruler.
• Zoom to selection: zooms to fill the spectrogram/waveform display with the current selection.
• Zoom to whole file: resets zoom time and frequency zoom levels to default.
• Zoom tool: When enabled, the zoom level will update to fit the selection to the window when
using the Time, Time-Frequency, or Frequency Selection Tools.

Zoom Sliders
 INTERACTIVE TOOLS 31

Magnitude Ruler Zoom Slider

The vertical slider in the lower right-hand corner of the spectrogram/waveform display controls
the zoom level for the spectrogram or waveform amplitude ruler(s). The buttons above the vertical
slider select which ruler it will affect when adjusted.

Time Ruler Zoom Slider

The horizontal slider in the lower right-hand corner of the spectrogram/waveform display controls
the zoom level of the time ruler.

Navigation Tools

Grab & Drag Tool

When zoomed in on an area, the Grab & Drag Tool [G] can be used to move through the time range
by clicking and dragging on the Spectrogram.

Dragging Rulers and Using the Mousewheel

The rulers to the right and below the Spectrogram display can clicked and dragged to reposition
the spectrogram and waveform to show a different time range, amplitude range or frequency
range.
In addition the range shown can be adjusted with your mouse:
▶ To adjust the zoom, place your cursor on top of a ruler and move the mouse wheel.
▶ To reset any of the rulers to the default display range, double click on the ruler.

NOTE
All active selections will be accurately preserved and scaled when zooming the main
display.

Channel Selectors
Channel Selectors appear on the left hand side of the spectrogram/waveform display when a file
is loaded. Channels can be enabled or disabled individually by clicking on the channel selector
buttons. Deselecting a channel will exclude it from playback, selections, and processing.
• If all channels are enabled, single-clicking on a channel selector will disable all other channels,
only the channel that was clicked on will remain enabled.
• When some of the channels are disabled, single-clicking on a disabled channel selector will
enable that channel without affecting the enabled disabled state of other channels.
• When any channel number of channels are disabled, double-clicking on any channel selector
will quickly enable all channels.
 INTERACTIVE TOOLS 32

KEYBOARD SHORTCUTS: STEREO FILE CHANNEL SELECTORS

Channel Order (Multichannel)

When working with multichannel files, the channel selector label order can be configured by
right-clicking on the time ruler or by clicking on the arrow to the right of the time format display.
The channel order options will update based on the number of channels in the active file tab. The
“Discrete” option will label all channels as “M” (mono).

Instant Process
Instant Process [I] is a selection tool modifier. Instant Process is only available in RX Standard and
Advanced.

When Instant Process is enabled, any new selection made will be immediately processed with the
selected module in the Instant PRocess menu. The module settings applied by instant process
reflects the current settings in the selected module.
When Instant Process is disabled, processing, editing and selection tools will function as they
normally do.

NOTE
If you hold Shift while using Instant Process, this will allow you to build up additional
selections. Once you release Shift, processing will occur. This is especially useful for
tools such as the Magic Wand, which will pick up additional harmonics upon second
click. If you don’t like that selection, also hold Alt, and start redrawing a new selection.
Release Shift once you’re ready to process.

Instant Process offers several different modes, accessible via a drop down menu, which will
instantly process the settings present in the named module/tab. The default settings are used,
but if you define custom settings in that module, Instant Process will recognize and apply those
custom settings. The modes are:

Instant Process: Attenuate

This mode will instantly apply the active settings from the Spectral Repair module’s Attenuate tab.
This is particularly useful if you see anything in the spectrogram you don’t wish to remove entirely,
but would rather quickly blend into the surrounding audio to make it less obvious or intrusive.

Instant Process: De-click

Applies the active settings from the De-click or Interpolate modules. De-click Instant Process will
automatically remove all clicks present in your selection, which is particularly useful for editing a
dialogue file, mismatching sample rate clicks and pops, and vinyl clicks.
 INTERACTIVE TOOLS 33

If you make a selection under 4000 samples in length, this mode will automatically use the
Interpolate module. Selection longer than 4000 samples will use the settings from the De-click
module. The De-click module is effective on selections above 4000 samples in size, as it is able to
identify clicks in relation to desirable audio, and then intelligently separate and remove the clicks.
If a selection is less than 4000 samples in length, it is likely a small selection of an individual click,
and Interpolate will fill the selection with audio information based on the surrounding audio.

Instant Process: Fade

This mode will instantly apply the active settings from the Fade module. This is particularly useful
if you’d like to smooth over a transition or edit point within a complex audio file, especially if it’s a
limited bandwidth selection, such as choosing to fade in a certain harmonic or audio event in an
audio file without changing the volume of the rest of the audio.

Instant Process: Gain

This mode will instantly apply the active settings from the Gain module. If you want to quickly
adjust certain audio events up or down in volume, you can simply paint over it to see the
immediate gain adjustment. For overall volume adjustment, use the Clip Gain line [Cmd+G /
Ctrl+G].

Instant Process: Replace

This mode will instantly apply the active settings from the Replace tab in the Spectral Repair
module. This is particularly useful if you see anything in the spectrogram you wish to remove
entirely, as it will use the audio information that surrounds your selection to instantly and
intelligently fill the gap.

Selection tools
The following selection tools are available in the RX Audio Editor:

Icon Name Description

Time selection tool Select a range of time within the file (horizontally within the
[T] spectrogram)

Time-Frequency Makes rectangular selections in the spectrogram display

Selection tool [R] to isolate sounds by time and frequency

Frequency Selection Makes frequency only selection (vertical in spectrogram)

tool [F]

Lasso Selection tool Makes a selection based on a free-form outline drawn with
[L] your cursor on the spectrogram

Brush Selection tool Draw a free-form selection using a defined brush size
[B] in time and frequency in RX’s spectrogram. The size of
the Brush Selection tool can be adjusted by clicking and
holding on the Brush Tool icon. Note: With the brush tool
selected, you can also hold Control/Command and move
the mouse wheel to make the brush size larger or smaller.
 INTERACTIVE TOOLS 34

Icon Name Description

Magic Wand Automatically selects similar harmonic content
Selection tool [W] surrounding the selected material. Click on the
spectrogram to select the most prominent tone under
the cursor when the magic wand is selected. Clicking
on an existing selection with the Magic Wand tool will
automatically select the overtone harmonics or related
audio components of your current audio selection. Note:
You can use the Brush or Lasso tools first to broadly define
a sound and then use Magic Wand to refine your selection
to include relevant harmonic material.
Harmonic Selection Duplicates your current selection to include harmonics
tool [Shift+Cmd+H / above it. Start by selecting the fundamental frequency
Shift+Ctrl+H] of audio with harmonics, then add or remove harmonic
selections with this tool before processing.

Selection Feathering Allows for crossfading of processed and unprocessed

tool [Shift+F] audio on both time and frequency axis. Adjust these
controls depending on how precise you want the edits to
be [Alt/Option+F]. Higher feathering values will introduce
longer crossfades, while lower values will create sharper
edits. Use the feather icon to enable/disable this feature.

When Feathering is disabled, the default time feathering of 20 ms remains on the

selection to prevent the introduction of unwanted sonic artifacts.

When feathering is enabled, there is potential for the selection visualization to lag
slightly.

View Clip Gain

Quickly enable/disable the clip gain curve overlay.
 INTERACTIVE TOOLS 35

Selection Modifiers

[Shift] - Add to selection

Hold down shift after making a selection in order to add another, separate selection. If any part of
the new selection overlaps any other, the selections will be grouped into one. This can be especially
powerful when combining several selection tools to create multiple selections of different size and
shape.

Multiple selections made using the Shift modifier key with different selection tools

[Alt/Option] - Subtract from Selection

Holding down Alt/Option will allow you use the currently chosen selection tool to remove or erase
any portion of an existing audio selection. This can be especially useful with the Lasso or Brush
tools, allowing you to edit or refine any piece of an existing selection.
This is also useful for refining complicated free-form selections. First make your lasso, brush, or
magic wand selection, and then hold alt while using the time, frequency, or time and frequency
tools to exclude entire time and frequency ranges from processing. Holding Alt/Option effectively
turns the Brush tool into a selection eraser for broad refinements and Lasso into a selection
“X-Acto knife” for detailed selection revision.

Selections refined to perfection using the Alt key modifier

 INTERACTIVE TOOLS 36

LINKING SELECTION STATE TO UNDO HISTORY EVENTS

Using Ctrl/Cmd-Z to undo any particular process will also bring back the previous
audio selection exactly as it was before applying any processing. In order to make use
of this feature, be sure that Store Selections with Undo History is enabled inside of
RX’s Preferences > Misc menu.

[Ctrl/Cmd] - Move Playhead without affecting Selection state

Hold down Ctrl/Cmd to move the transport’s playhead to any position without erasing your current
audio selections. This can be especially useful with previewing or comparing complex audio
selections without having to remake these specific audio selections.

[Mouse Over] - Grab and Drag Selection

After using any of RX’s tools to select a portion of your audio, when the mouse is subsequently
placed on top of any selection, a Grab and Drag hand cursor will be displayed automatically,
allowing you to change the position of that selection.
 UNDO HISTORY 37

9. Undo History

Overview
In addition to the Undo and Redo commands, the Undo History list allows you to see a timeline of
changes you’ve made and non-destructively revert back to earlier states. RX keeps a log of all your
edits in this Undo History. When modifying Clip Gain or processing with the Module Chain related
edits will be added under parent items, named “Clip Gain” and “Module Chain” respectively, in the
Undo History list.

NOTE
You can rename items in the undo history by double-clicking them. You can save
an .rxdoc of your current file to retain the undo history list, you can find more
information on RX documents in the Working with Files chapter.

Export History
You can use the Export History feature to save an XML file, listing the entire undo history for your
particular file. For forensic and archival purposes, it is often useful to have an official record of all
edits that were made to a particular file. When a file’s history is exported, the following information
will be stored in an XML file:
• RX Version Number
• Time and Date
• Corrected File
• Number of Channels
• Sampling Rate/Bit Depth
• Edit History: Parameters and Selections

Restore Selection
You can now restore selected regions to earlier states in the Undo History. When hovering your
cursor over an item in the History Pane with a selection made in the spectrogram, an icon appears
in line with the text. Clicking this icon will restore your selection to that point in history.
 UNDO HISTORY 38

This is the same as going back to that point in undo history, except only for your current selection.

Expandable History List

The History list is now expandable, so you can see more of your edit history without scrolling.
 APPLICATION MENUS 39

10. Application Menus

File
The file menu provides options for creating, importing and exporting files. For information about
file management in the RX Audio Editor, see the Working with Files chapter.

Edit
The Edit menu includes the following options:
• Undo [Ctrl/Cmd-Z] Reverses the last action taken.
• Redo [Ctrl/Cmd-Shift-Z; Ctrl/Cmd-Y] Cancels the undo.

UNDO HISTORY
The RX Audio Editor includes an Undo History event list in each file tab. The Undo
History list stores a list of all processing or edit operations applied to a given file.

• Cut[Ctrl/Cmd-X] Removes the currently selected audio and stores it temporarily on the
Clipboard.
• Copy [Ctrl/Cmd-C] Makes a copy of the currently selected audio and places it on the Clipboard.
• Paste[Ctrl/Cmd-V] Places audio that has been copied or cut to the Clipboard at the current
cursor point.
• Paste Special Provides additional options for placing the Clipboard data:
• Insert [Ctrl/Cmd-Alt/Opt-V]: Inserts the audio from the Clipboard and moves audio in the
project [does not overwrite]
• Replace [Ctrl/Cmd-Alt/Opt-Shift-V]: Replaces audio in the project with audio from the
Clipboard
• Mix [Shift-V] Combines the audio from the Clipboard with audio in the project
• Invert and Mix [Alt/Opt-V] Inverts the audio in the Clipboard and then mixes it with audio in
the project. This is useful when you want to compute the difference between two signals.
• To Selection [Alt/Opt-Shift-V] Pastes audio from the clipboard only within the selection
bounds, regardless of the copied audio’s length. If the copied audio is longer than the new
selection, the audio will be cropped to fit. If the selection is longer than the audio being
pasted, silence will be inserted to fill the remaining space.
• To Clip Gain Only [Ctrl/Cmd-Shift-V] Pastes only clip gain information to the current
selection
• Deselect [Ctrl/Cmd-D] If audio is selected, deselects it and places the anchor sample at the
start of the selection.
• Reselect [Esc] Restores the last selection if you have no current selection.
• Select All [Ctrl/Cmd-Shift-D] Selects the entire open file.
• Invert Selection [Ctrl/Cmd-Shift-I] Selects everything that isn’t currently selected.
• Invert Selection Frequencies [Ctrl/Cmd-I] Selects everything in the current time range that isn’t
selected. This is useful for refining processing by first selecting what you don’t want to process,
then inverting the selection frequency.
 APPLICATION MENUS 40

• Select Harmonics [Ctrl/Cmd-Shift-H] Refines the current selection to include more harmonics.
For this feature to work well, try it with a simple selection that includes only the fundamental
harmonic of what you are trying to select. You can also use the Magic Wand tool to
automatically refine a selection to include the appropriate harmonics.
• Begin Selection at Playhead [Left Bracket] during playback only If audio is currently selected,
this will automatically adjust the selection to begin at the current playback position.
• End Selection at Playhead [Right Bracket] during playback only Automatically create a
selection between the current playhead position and the original anchor playhead position.
• Delete Selection [Delete on a time selection] Deletes the selected audio and closes the space
with audio from either side of the timeline.
• Silence [Delete on a frequency, time-frequency, or freeform selection] Deletes selected audio
and replaces it with silence.
• Trim to Selection [Ctrl/Cmd-T] Deletes all audio except for the selected audio.
• Add Marker or Region [M] This will create a new marker point at the current location of the
cursor/playhead or create a new region if any audio is selected.
• Edit Cursor Mode Changes the behavior of the editor cursor to select by time and/or frequency,
or to zoom. These modes can also be selected from the Cursor Mode buttons.
• Select Time [T] Makes a time selection
• Select Time/Freq [R] Makes a Rectangular time-frequency selection
• Select Freq [F] Makes a frequency selection for the duration of a file
• Lasso [L] Selects everything in a freely defined area
• Selection Brush [B] Selects everything in a predefined radius
• Selection Wand [W] Intelligently selects material similar to whatever is under your cursor
[magic wand]
• Zoom Time [Z] Zooms in time
• Zoom Time/Freq [Shift-Z] Zooms in time and frequency
• Zoom Freq [Alt/Opt-Z] Zooms in frequency
• Grab Time [G] Grabs and drags the view in time
• Grab Time/Freq [Shift-G] Grabs and drags the view in time and frequency
• Grab Freq [Alt/Opt-G] Grabs and drags the view in frequency
• Snap [Ctrl/Cmd-Shift-;] Snap selections to the boundaries selected in the “Snap To” sub-menu.
• Snap to:
• Markers
• Ruler Coarse
• Ruler Fine
• Zero Crossings
• All
• None

View
The View menu includes the following options:
 APPLICATION MENUS 41

• Collapse/Expand Module Panel Collapses the module list panel of RX into a row of icons.
• Time Format RX’s time scale and playhead location counter can be set to show different time
units. Learn more about changing the time format in the Transport chapter
• Follow Playhead [Ctrl/Cmd+P]: Toggles whether or not the current view follows the playhead
position during playback.
• In Page mode, the view will follow the playhead one view length at a time.
• In Continuous mode, the view is centered on the playhead as it moves across the file.
• Effect Overlays: This sub-menu allows you to turn special display features for the De-clip and
Spectral Repair modules on and off. For an overlay to be visible, you need to have the option
selected in the view menu, and the corresponding module window needs to be open.
• De-clip Threshold When the waveform is visible, the threshold settings and controls appear
as white lines within the display. This display can be used to adjust the de-clip threshold
settings.
• Spectral Repair Source Regions Displays effective region bounds when using the Spectral
Repair module.
• Clip Gain: Enables/disables the Clip Gain envelope in the main editor window. The Clip Gain
curve can also be toggled on or off by clicking the “View Clip Gain” button to the right of the
selection tool buttons, or by using the following keyboard shortcuts: Command+G (Mac); Ctrl+G
(Windows).
• Show Channels Separately: Toggles how channels are displayed in the spectrogram/waveform
view. When enabled, each channel will be drawn in its own lane in the main editor view.
When disabled, the channels are drawn in one summed view. This allows for greater vertical
resolution, especially when working with multichannel files. This option can also be toggled on
or off by: clicking the Channel View button to the left of the mini waveform overview display or
by using the following keyboard shortcuts: Command+Shift+C (Mac); Ctrl+Shift+C (Windows).
• Channel Mode: Selects between Left/Right and Mid/Side channel mode. This affects the
waveform and spectrogram display as well as any processing that may be applied. This option
is only available if Show Channels Separately is enabled.

NOTE
Music Rebalance and Dialogue Isolate are not designed for use in Mid/Side mode.

• Spectrogram Settings: Opens the Spectrogram Settings window.

 APPLICATION MENUS 42

• Enter full screen: Enables full screen mode.

Modules
The Modules menu includes options for opening module windows, running Learn on the current
selection without opening the module window, and rendering module settings on the current
selection without opening the module window.

Open Module Window

Open a module window by selecting it from this menu.

Learn
Run a Learn pass on the current selection without opening the associated module window.
 APPLICATION MENUS 43

Render
Render settings for any module on the current selection without opening the associated module
window.

Transport
The Transport menu includes the following options:
 APPLICATION MENUS 44

• Input Monitor [Alt/Option-I] Enables input monitoring. When input monitoring is enabled, the
input signal of RX will be routed to the output signal of RX.
• Arm for Recording / Record / Stop Recording [Alt/Option-Space] Runs the next possible step
for recording. If you have not opened a new file, Arm for Recording will open the New File
dialogue box for you.
• Rewind [Return] Sets the playhead to the beginning of the file.
• Play/Stop [Space] Starts or stops playback. If Input Monitoring is enabled, starting playback
will temporarily suspend Input Monitoring.
• Loop Playback [Control/Command-L] Toggles playback looping. If nothing in the file is
selected, the end of the file will loop back to the beginning.
• Playhead Follows Playback [Control/Command-R]: Toggles the behavior of the playhead on
stop. If this is enabled, the playhead will return to the anchor sample (the position before
playback began).
• This is useful for comparing processing. If this is disabled, the anchor sample will be set
to the current playhead position. This is useful for moving through a file while listening for
irregularities.

Window
The Window menu includes the following options:

• Batch Processing [Ctrl/Cmd+B]: This gives you access to file based batch processing, as
explained in the Batch Processor chapter.
• Waveform Statistics [Alt/Opt+D]: This gives you access to informational readouts on a variety
of amplitude measurements, as explained in the Waveform Statistics chapter.
• Spectrum Analyzer: The Spectrum Analyzer displays an analytical view of your audio. More
information is located in the Spectrum Analyzer chapter.
• Markers and Regions [Alt+M]: Markers and regions allows you to define and save particular
points or selections in time for your audio file. More information is located in the Markers and
Regions
• Close All Floating Windows/Reopen Closed Windows [Ctrl/Cmd+Opt+W]: Closes or reopens
all floating windows.
• File Info: View metadata and other information about the audio file, as explained in the Working
with Files chapter.
 APPLICATION MENUS 45

• Next File (Control-Tab) Changes RX’s current file tab to the next file in the window order.
• Previous File (Control-Shift-Tab) Changes RX’s current file tab to the previous file in the
window order.

Help
The Help menu includes the following options:
• Launch Help: Open the product help documentation 8 Keyboard Shortcuts: Open the Keyboard
Shortcut Guide
• Getting Started: Launch the First Time User Tour
• What’s New See the new features and tools RX 11
• Online Video Tutorials: View articles and tutorials on iZotope.com
• Give Feedback: Send feedback directly to the iZotope team
 PREFERENCES 46

11. Preferences

Audio

• Driver Type: Selects the audio device driver type (for example: ASIO, CoreAudio, or RX Monitor)

Some hardware devices monopolize the audio drivers when sending audio clips to
RX via RX Connect. If you are not able to hear the audio sent to RX from your DAW
with RX Connect, instantiate RX Monitor on a track in your session and change the
audio driver to RX Monitor in the Driver type menu.

• Input/Output Device: Choose the device/sound card you want RX to use for playback and
recording.
• Device Refresh: Rescan for newly connected audio devices.
• Buffer Size: The total playback buffer size. In general, lowering these buffer sizes will improve
meter responsiveness and lower latency, but increase CPU needs. Raising buffer sizes will
lower CPU cost but increase latency. It’s worth exploring these ranges to find values that work
best on your system.
• Num Buffers: Number of playback sub-buffers. (Windows MME Only.)
• Composite View gain reduction: Nondestructively reduces the output gain of all clips included
in the Composite View tab by the amount specified in the dropdown.
• Channel Routing: Opens the Channel Routing window. Input and output channel routing can
be configured in this dialog when working with ASIO/CoreAudio driver types or to configure
multichannel output routing.
• Preferred Layout (Multichannel only): Selects the default channel ordering option when
multichannel files are loaded in the application. ADV
• Stereo Downmix (Multichannel only): Selects the stereo downmix configuration to use when
monitoring multichannel files with a stereo audio output device. ADV

RX does not support downmixing or upmixing files when saving or exporting. The
downmix option only applies to playback of multichannel files on stereo systems.
 PREFERENCES 47

• Configure Driver: Launches the manufacturer’s driver configuration dialog.

• Release when not in use: Auto-closes the audio device when playback in RX stops, freeing it for
use in other audio applications. Disable this if playback from RX isn’t responsive enough.
• Test Tone: The test tone generator is useful for testing your speakers, audio hardware and
listening environment. Tones at set frequencies or at a custom frequency can be used as test
tones, as can white or pink noise. In addition, a Channel Identification mode will identify left and
right speakers.
• Enable: Starts playback of a test tone.
• Type: Sets the type of test tone to play.
• Volume: Sets the volume of the test tone.
• Frequency: Sets the frequency of the test tone.
• Output Gain: Output gain allows you to nondestructively adjust the playback level of RX 11
Audio Editor.

Display

• Show tooltips: When enabled, hovering over an RX feature with the mouse cursor will show a
short description of the feature.
• Display cursor coordinates in status bar: When enabled, the time coordinate of the cursor is
shown in the status bar at the bottom of the RX main window. The amplitude of the audio at the
cursor position and the frequency at the cursor position is also shown.
• Show analog waveform: When digital audio is played back, it is converted to analog. The peak
values in the analog waveform can be larger than the peaks in the digital waveform, leading to
clipping in the output of a digital-to-analog converter. When Show analog waveform is enabled,
RX will compute an analog waveform in the background. Any peaks will be highlighted in red on
top of the existing digital waveform.

NOTE
RX will automatically display an analog waveform when zooming in at extreme
zoom levels.

• Offload waveform calculations: When enabled, RX’s waveform display will be computed in
the background. This allows very large files to be loaded very quickly, but it slows down RX’s
waveform displays.
 PREFERENCES 48

• Waveform interpolation order: If you zoom into the waveform so that individual samples
become visible, RX will display an upsampled analog waveform as well as the individual digital
samples. The interpolation order controls the quality of upsampling. Higher values yield more
accurate analog waveforms at the expense of CPU usage.
• Brightness: Adjusts the general brightness of the RX interface, allowing you to make RX more
readable on your specific display.
• Floating window opacity: Changes the opacity for RX’s floating windows. This can be useful
if you wish to leave floating windows on top of the spectrogram and waveform without
completely obscuring the display.

Keyboard

CUSTOMIZING KEYBOARD SHORTCUTS

While RX includes default keyboard shortcuts, you can also customize them to
your liking. Refer to the Keyboard Shortcut Guide to reference a list of default key
commands and internal shortcut command names used by RX. Referring to the guide
can help you quickly identify the name of the key commands you want to customize
and search for them in the “Show commands containing” field (explained below.)

• Presets: Save groups of key assignments with this tool.

• Show commands containing: Lets you search by keyword for a command you want to assign
to a keystroke.
• Shortcuts for selected command: Shows if there are any keystrokes assigned to the command
selected in the above menu.
• Remove: Removes the currently assigned keystroke from a command.
• Press Shortcut Key: To assign a new keystroke to a command, select the command from the
menu, then click in this field and press a key or combination of keys.
• Assign: Assigns the entered keystroke to the current command. The shortcut will only be
assigned to the current command if you press this button.
 PREFERENCES 49

• Shortcut key currently used by: Lists commands that the current keystroke is assigned to.

USING THE ALT MODIFIER ON WINDOWS

On Windows systems, by default, “Alt + a letter” will open the corresponding menu
for your currently open application. Alt + V for example will open RX’s View menu
drop down. By default, none of RX’s shortcuts should conflict with these keyboard
shortcuts, however if you wish to assign Alt + V to another operation, it will take
precedence over the View menu.

Misc

• Session data folder: Allows you to choose a different folder to save RX’s temporary session
data. These files are created to allow actions to be undone and sessions to be recalled in RX.
Because these can be very large, it is best to set this to the drive on your computer with the
most free space.
• Time scale frame rate: This sets the frame rate used to draw the time scale when RX is set
to display the time code (see View menu or right-click the time ruler to change this setting).
Choose from a list of standard frame rates or click in the combo box to define a custom frame
rate.
• Default full-bandwidth paste mode: This controls RX’s behavior when pasting a full-bandwidth
audio selection. Insert will move aside existing audio, Replace will overwrite existing audio, and
Mix will add to existing audio.
• Default limited-bandwidth paste mode: Similar to the full-bandwidth paste mode, this controls
RX’s behavior when pasting a limited-bandwidth audio selected.
• Reopen previous audio files when app starts: When enabled, RX will open all of the files
(including edits, processing and undo history events) that were present when RX was last
closed. Disabling this option will open the RX Audio Editor in its default state (no files loaded.)
• Reopen previous floating windows when app starts: When enabled, any floating windows
that are open when the application is closed will be reopened the next time the application is
opened.
• Automatically open files ending with .L and .R as split stereo: Mono audio files with (.L
and .R) as well as (.1 and .2) extensions will be opened as stereo files when this option is
enabled.
 PREFERENCES 50

• Recall selections during undo/redo: When this is enabled, RX will recall the selection used for
an item in the undo history. When stepping through the undo history events, selections that
were used for each event will be restored along with the audio.

DISABLING THE SELECTION UNDO/REDO OPTION

Sometimes it is useful to turn this off if you need to compare undo history items
and not break your current selection (like a useful loop).

• Play only selected channels: If only a single channel of audio is selected and this option is
enabled, all other channels will be muted during playback.
• Calculate RMS using AES-17: Uses the AES-17 1998 standard for RMS calculations (0 dB is
a full scale sine wave) in the level meter, Waveform Statistics and Leveler modules. The other
option is when 0 dB is the RMS of a full-scale square wave. These options differ by 3 dB.
• Pre- and Post-Roll during preview (ms): When Previewing audio processing in any module,
the specified time amount will be added to the beginning and end of the previewed selection in
order to provide contrast between unprocessed and preview-processed audio.
• True Peak Calculation: RX provides several modes for measuring true peak levels in Waveform
Statistics:
• BS.1770 baseline mode uses a “minimal” 4× oversampling with the specific filter from the
BS.1770 standard.
• HQ oversampling mode uses a higher (at least 9×) oversampling factor with a flat passband
response. It does not exhibit the passband ripples of the BS.1770 filter, but may sometimes
produce lower readings than the baseline filter.
• Maximally compliant mode runs both of the above measurements and takes the higher of
two detected true peak levels. In this way, it achieves compliance with most existing true
peak measurement software.
• Max Processing Threads: When RX is applying resource-intensive processing it can slow down
your computer. At the expense of speed, adjusting the Max Processing Threads to a lower
value limits the memory resources RX consumes in order to have a smoother experience while
doing other tasks as RX runs in the background. This is especially useful when using the Batch
Processor. The default value is ‘Auto’, and the minimum value is 1 thread.

License and Updates

• Authorization: Provides options to authorize or de-authorize RX (explained in the Authorization

chapter).
 PREFERENCES 51

• Help iZotope improve RX by sending anonymous usage data: Enables sending anonymous
usage data to help us improve our products.

Plug-ins
RX 11 Audio Editor supports the use of the following plug-in formats in the “Plug-in” module:
• VST2: Windows and Mac (Intel and Rosetta only)
• VST3: Windows and Mac
• AU (AudioUnit): Mac Only

UPDATE YOUR PLUG-INS

Please make sure your third-party plug-ins are updated to their latest version in
order to ensure that RX will be able to scan them correctly. Contact your plug-in
manufacturers for updated installers, if necessary.

ABOUT THE ABOVE SCREENSHOT

Only Windows users and Mac users working on Intel-based systems (or Silicon-based
systems running Rosetta) will see the option for VST2. For all other systems, this
format is unsupported and will not appear as an option.

• Plug-in Lists: Displays plug-ins that have been scanned for use in the “Plug-in” hosting module
of the RX Editor.
• Enable: Enables that plug-in format for use in the RX Audio Editor. This will trigger plug-in
scanning to begin in the background.
• Disable: Disables the associated plug-in format. This will clear the scanned plug-in list for that
format. Re-enabling that plug-in format will prompt RX to re-scan that plug-in format.

NOTE
If a plug-in failed scanning for any reason, the plug-in’s name will be prefixed with
an error tag (ex: [Crashed] or [Failed]) to help troubleshoot the failure.
 PREFERENCES 52

• VST plug-in folders: Allows you to add or remove custom VST2 plug-in folder paths. RX uses
the system VST2 plug-in folder by default. If you are using a custom directory for VST2 plug-ins,
use this option to ensure that those VST2 plug-ins will be scanned.

NOTE ABOUT SUB-FOLDERS WHEN SCANNING FOR PLUG-INS

RX will scan the first level of sub-folders in the custom VST2 folder. If some of your
plug-ins do not show up when you scan them, and you know they’re in a subfolder
of your plug-in folder, try moving them up one directory level.

• Group plug-ins by name in plug-in menus: When enabled, the RX plug-in menu will group
plug-ins by common first words, usually the manufacturer’s name. When disabled, the RX
plug-in menu will appear as a single, alphabetically sorted list.
• Rescan: If RX detects that a plug-in is unstable, it will blacklist it and prevent it from being
opened. The rescan option allows you to clear the blacklist of unsupported plug-ins and rescan
all installed plug-ins in case an RX update or an update from the plug-in manufacturer resolves
the issue.

Assistant

• Enable Repair Assistant analysis and processing suggestions for: Informs what common
audio problems will be considered when Repair Assistant analyzes a selection. When an option
is checked in this list, the associated audio problem will be included in analysis and processing
suggestions. Options include:
• Clipping
• Clicks
• Hum
• Noise
• Reverb
• Ess
 COMPOSITE VIEW 53

12. Composite View

STD & ADV

Overview
The Composite View feature in the RX Audio Editor combines all active tabs into one “Composite”
tab that allows you to apply the same processing to multiple files simultaneously. Composite
View can be a valuable tool for increasing efficiency when performing repetitive spectral editing
functions.

Workflow
To enter Composite View mode, click on the Enter Composite View button.

When you are done making changes in Composite View, click on the Exit Composite View button to
continue working with your individual tracks.

Important Notes
• Composite View is designed to function as a bulk editor, it is not intended for mixing. A
maximum of 32 files can be collapsed into a Composite view tab.
• Composite View assumes that all files start at the same point.

USE THE SIGNAL GENERATOR MODULE TO INSERT SILENCE

If needed, you can adjust file start times by inserting Silence with the Signal
Generator or using edit commands to modify timing before collapsing your files
into a Composite View tab.

• Composite View learns on the first track only.

• Sample Rates: Composite View requires all files to have matching sample rates.

RESAMPLE FILES USING THE RESAMPLE MODULE

Use the Resample module to conform each file tab to the same sample rate value.

• Spectrogram Display: Composite View displays a summed composite spectrogram/waveform

display for all files collapsed into the Composite tab. To view individual file displays, simply exit
Composite View.
• Editing and Processing: All processing and edits applied in Composite View will be applied to
all tracks.
 COMPOSITE VIEW 54

• Undo: To revert edits made in Composite View on a subset of tracks you can exit composite
view and use the Undo History list to undo changes made to individual files.
• Playback: Composite View plays back the sum of all files included in the Composite tab.
• Channel selection states in individual tabs are not respected in Composite View playback or
processing.
• Markers and Regions are not supported in Composite View.
• Compare Settings functionality is not supported in Composite View.
• De-bleed is not recommended for use in Composite View.
• Wow & Flutter will not work in Composite View.
• Loudness Control will not work in Composite View.
 BATCH PROCESSOR 55

13. Batch Processor

Overview
The Batch Processor allows you to process multiple files at once with a custom module chain,
streamlining time-consuming tasks. The Batch Processor can run simultaneously with the RX
Audio Editor, allowing you to process in bulk while working in RX.
The Batch Processor window can be opened using the following methods:
• Navigate to the RX Audio Editor “Window” menu and select “Batch Processor”.
• Open the Batch Processor window using the default keyboard shortcuts: Cmd+B (Mac); Ctrl+B
(Windows).

Batch Processor Presets

Click here to save, load, or update a Batch Processor preset. Output options and module chain
contents can be saved together as presets.
Learn more about saving presets
More information about saving presets is covered in the Common RX Module Controls chapter.

Input
The Input section allows for files to be added for processing. Added files will display the channel
count, bit depth, sample rate and file length.

Adding Files
Files can be added to the Batch Processor window by using one of the following methods: Click the
Add files button at the top of the Input section. Drag and drop files or folders from Finder (Mac) or
Windows Explorer (Windows) into the Input files section of the Batch Processor window. Folders
that are added will retain their folder structure in the output directory.
 BATCH PROCESSOR 56

TIP
Add multiple files from the same folder by holding the Command key (Mac) or ctrl key
(Windows) when selecting files from the Finder or Windows Explorer dialog.

Removing files

Individual files can be removed from the Batch Processor by clicking the ‘X’ that appears when
hovering over a file. Clicking the Remove all files button at the top will clear all files from the input
section.

Module Chain
Processing steps can be configured in the Module Chain area of the Batch Processor window. You
can create a new Module Chain or recall a Module Chain preset by selecting it from the dropdown
menu.
Learn more about the Module Chain
See the Module Chain chapter for more information about the controls available in the Module
Chain.

NOTE
Wow & Flutter is not available in the Batch Processor.

Output
The Output Section includes options for selecting the location of the processed files, modifying file
names, and setting the output file format.
Use the “Choose folder…” button to select the destination for the processed files from the system
dialogue that appears. To send files to the same folder they are sourced from, click the ellipsis
button in the output panel and choose ‘Same as Original Location’.
 BATCH PROCESSOR 57

Input folder structure will be retained in the output location of your choice.

Naming
To modify a file name, add the desired text to the naming field and select the Before “|<-” or After
“->|” icons.

File Format
Choose the output file format and configure the options associated with the selected file format
(e.g. bit rate, quality, etc…). Available File Formats:
• WAVE
• AIFF
• FLAC
• Ogg Vorbis
• MP3
 REPAIR ASSISTANT 58

14. Repair Assistant

Overview
Repair Assistant is designed to detect common audio problems in a recording and quickly
provide suggested settings for improving them. The Repair Assistant interface provides access
to parameters that allow the user to fine-tune the suggestions to their own liking. Repair Assistant
is available in the RX Audio Editor as a module and as a plug-in.

Repair Assistant Module

To open Repair Assistant in the RX Audio Editor application, simply click the Repair Assistant
button in the upper right-hand corner of the RX Audio Editor window.

After opening the module, choose between Voice and Music modes. This changes the type of
processing chains generated and is used in analysis to measure noise amounts and determine
processing components and settings. Voice mode is intended for spoken word or dialogue. Music
mode is intended for a full-mix music recording, individual instruments, or isolated sung vocals.
Next, click Learn to analyze your audio. Repair Assistant is designed to be trained on audio
that contains dialogue, vocals, music, and sound effects. It isn’t necessary to learn on noise-only
segments of audio like with other modules such as Spectral De-noise. Repair Assistant analyzes
any selection made in the Spectrogram (if no selection is made it analyzes as much data as
necessary in order to make an effective suggestion) and automatically adjusts the settings for
optimal repair of your audio. After analysis is complete, you will be able to adjust the various
parameters of the tools that are working inside the assistant.
To apply Repair Assistant’s recommended processing to the current selection, select the
processing option in the Repair Assistant window and click the Render button in the bottom
right-hand corner of the window.
 REPAIR ASSISTANT 59

The Open Module Chain allows you to save Repair Assistant’s recommendations as a Module
Chain preset, or make changes to the generated settings before rendering. To open Repair
Assistant’s recommendations in the Module Chain, click the Open Module Chain button in the
Repair Assistant footer area (this feature is available in RX 11 Advanced only).

Repair Assistant Plug-in

The experience in the Repair Assistant Plug-in is very similar to that of the Audio Editor, but the
plug-in analyzes your audio as you play audio, as opposed to analyzing the entire file/selection. To
use the plug-in, simply choose your Mode (Voice or Music), click Learn, and play your track. Keep
playing until Repair Assistant has finished listening to your audio and created its suggestion.

When using AudioSuite, clips must be longer than 7 seconds. Learning a clip less than
7 seconds long will cause the plug-in to 'learn' indefinitely and appear to hang in this
state. When using AAX, you may loop shorter clips in order to run learning.

Voice Mode

Common Controls
• The On/Off button in the upper left corner of each section enables/bypasses that section.

• The Audition button in the upper right corner of each section allows you to hear only what is
being removed by that section.
 REPAIR ASSISTANT 60

Clean Up
• De-noise: Controls the applied amount of Dialogue Isolation processing.
• De-reverb: Controls the applied amount of De-reverb processing.
• De-hum (on/off): Activates/deactivates De-hum processing (De-hum will only operate if hum is
detected during learning).
• De-click (on/off): Activates/deactivates De-click processing.

Tone
• %: Controls the amount of boosting/cutting bi-directionally (based on EQ shape).
• % (Tilt): Positive boosts the high end and lowers the low end; Negative boosts the low end and
lowers the high end.
• % (Scoop): Positive boosts the high end, boosts the low end, and lowers the mid range;
Negative lowers the high end, lowers the low end, and keeps the mid range as-is.
• Tilt/Scoop: Choose which EQ shape to use.
• Lo-cut (on/off): Activates/deactivates lo-cut filter.
• Hi-cut (on/off): Activates/deactivates hi-cut filter.

De-Ess
• Shaping: Determines how much the spectral shape of the sibilant is changed.
• A setting of 0% leaves the natural shape of the sibilance.
• A setting of 100% flattens the shape of the sibilant toward a specified spectral profile.
• Threshold: Determines the level at which the De-ess module begins compressing sibilance.
• Fast/Slow: Sets the attack and release times for the processing.

De-Clip
• Threshold: Defines the level used for detection of clipped intervals. Generally, this should be set
just below the actual level of clipping. To set the threshold, move the Threshold slider until it
lines up with the place in the histogram just below where clipping is concentrated.
• Output Gain: Selects the gain to be applied to the selection after De-clip.
• Limiter (on/off): Applies a true peak limiter after processing to prevent the processed signal
from exceeding 0 dBFS.
 REPAIR ASSISTANT 61

Music Mode

Common Controls
• The On/Off button in the upper left corner of each section enables/bypasses that section.

• The Audition button in the upper right corner of each section allows you to hear only what is
being removed by that section.

Clean Up
• De-noise: Controls the applied amount of Adaptive De-noising (best for eliminating room tone
in vocals or mic’d acoustic instrument recordings).
• De-squeak: Controls the amount of processing for removing fret and finger noises/squeaks
that occur when sliding up or down the guitar neck between notes or chords.
• De-hum (on/off): Activates/deactivates De-hum processing (De-hum will only operate if hum is
detected during Learning).
• De-Pick (on/off): Activates/deactivates De-pick processing

Tone
• % (Tilt): Positive boosts the high end and lowers the low end; Negative boosts the low end and
lowers the high end.
• % (Scoop): Positive boosts the high end, boosts the low end, and lowers the mid range;
Negative lowers the high end, lowers the low end, and keeps the mid range as-is.
• Tilt/Scoop: Choose which EQ shape to use.
 REPAIR ASSISTANT 62

• Lo-cut (on/off): Activates/deactivates lo-cut filter.

• Hi-cut (on/off): Activates/deactivates hi-cut filter.

De-Harsh
• Cutoff: Specifies the crossover point between musical signal (to be preserved) and “harshness”
(to be reduced). The Cutoff frequency value functions as the lower boundary for harshness
detection.
• Threshold: Determines the level at which the De-harsh module begins compressing harshness.
• Fast/Slow: Sets the attack and release times for the processing.

De-Clip
• Threshold: Defines the level used for detection of clipped intervals. Generally, this should be set
just below the actual level of clipping. To set the threshold, move the Threshold slider until it
lines up with the place in the histogram just below where clipping is concentrated.
• Output Gain: Selects the gain to be applied to the selection after De-clip.
• Limiter (on/off): Applies a true peak limiter after processing to prevent the processed signal
from exceeding 0 dBFS.

Repair Assistant Preferences

To bypass analysis by any of the Repair Assistant modules, click on the gear button in the upper
right-hand corner of the Repair Assistant window to open the Assistant Preferences. All modules
are selected by default, and deselecting any module will remove it from Repair Assistant’s analysis.

These preferences apply only to Repair Assistant in the RX 11 Audio Editor application, not the
plug-in. For more information about the RX Preferences menu, see the Preferences chapter.
 COMMON MODULE CONTROLS 63

15. Common Module Controls

Overview
The RX Audio Editor is designed to give you a range of processing options. Most of the modules
in RX feature multiple processing modes, ranging from fast algorithms that sound great on most
material to very time-intensive algorithms for critical applications. Understanding the Presets,
Preview and Compare controls will help you save time, especially when taking advantage of RX’s
more powerful processing modes.

Presets

Each module in the RX Audio Editor features a preset menu that allows you to choose between
factory presets and custom presets that you have saved. Any preset saved in a module in the RX
Audio Editor can be opened in the corresponding RX plug-in, when applicable.

• Add Preset: Creates a new preset.

• Remove Preset: Removes a preset from the drop-down list.
• Rename Preset: Changes the name of a preset.
• Set Preset Shortcut: Allows you to define a keyboard shortcut for any preset in order to recall
and apply different module settings quickly.
• Import Preset: Allows you to import presets (from another machine or another user, for
example).
• Reload Preset: Rescans your preset directory for this module in order to refresh the available
preset list.
• Explore Preset/Reveal Presets in Finder: Opens Windows Explorer or Finder window to the
location of your presets on disk.

Preset directory locations

• Windows factory preset install location: C:\Program Files\iZotope\RX Pro Audio Editor\Presets\
• Windows user preset save location: C:\users\username\Documents\iZotope\RX Audio
Editor\User Presets\
• Mac factory preset install location: /Library/Application Support/iZotope/RX Pro Audio Editor/
Presets/
• Mac user preset save location: ~/Documents/iZotope/RX Audio Editor/User Presets/
 COMMON MODULE CONTROLS 64

If you previously installed RX Pro, then your presets folders will be in these locations:
• Windows user preset save location (RX Pro previously
installed): C:\users\username\Documents\iZotope\RX Pro Audio Editor\Presets\
• Mac user preset save location (RX Pro previously installed): ~/Documents/iZotope/RX Pro
Audio Editor/Presets/

Preset Dirty Flag

Making changes to a preset will add an asterisk [ * ] to the preset name to show that it has been
modified.

You can use Update Preset to commit your changes, or Add Preset to create another preset with
your new settings.

Module Footer Controls

Preview
Many RX modules feature a Preview button in the bottom panel of the module window. Some
modules don’t support Preview because of the time-intensive nature of their processing, in many of
these cases the Compare settings option is available (explained below).
Preview allows you to make adjustments to controls and hear the results without the need to
process and undo processing multiple times to achieve your desired results.
Preview will apply to your active selection or if no selection is made, Preview playback will start
from the current playhead position. When Loop is enabled in the Transport, Preview playback will
loop.
• Preview [Shift-Space]: Plays a pre-rendered preview of the module’s current settings on the
selected audio. During preview, module settings can be adjusted and adjustments will be heard
within the length of the current preview buffer. For most modules, the Preview buffer size is
about a half of a second long, but the Preview Buffer Length can be adjusted by accessing the
Preview Options, explained below.

• Bypass [Shift-B] Bypasses module processing during preview.

 COMMON MODULE CONTROLS 65

• Preview Options [+]: Allows adjustment of the Preview Buffer size.

For more CPU-intensive settings, like the highest quality algorithms in Spectral De-noise and the
highest quality De-click settings, RX can buffer playback to allow you to preview these slower than
real-time processes.

NOTE ABOUT PREVIEW BUFFERING

When Previewing module processing, the active buffer length for preview rendering is
tinted red in spectrogram/waveform display.

Pre- and Post-Roll

When previewing an effect, it is often very helpful to hear a small portion of the unprocessed
audio before hearing your processed selection. This can provide a much clearer comparison and
allow you to more easily discern whether or not the current processing settings are producing the
desired effect.
By default, RX will play back one second of unprocessed audio before and after the current
selection when previewing your processing. The Pre- and Post-roll times can be defined in the
Preferences > Misc window. RX can play up to ten seconds of audio before or after the previewed
selection. Pre- and Post-Roll will also occur when previewing a looped region of audio

HOW TO DISABLE PRE- OR POST-ROLL

Set the Pre- and Post-roll times in the Misc tab to 0.

TIP
Pre- and Post-roll can also be simulated manually by holding Control (Windows) or
Command (Mac) to set the playhead to any desired position while preserving your
audio selection. Once the playhead is set, clicking on Preview in the desired module
will then start the Preview playback from the playhead position.

Compare
When you want to quickly try a lot of different settings, use the Compare feature. In some cases,
you might not know what settings of a module will give you the best, most transparent results. By
hitting the Compare button instead of the Process button, you can audition multiple settings of the
same module and then audition the results side by side in the Compare Settings.
While one group of settings is processing in the background, you can return to the module and try
a different group of settings. Learning to use the Compare Settings tool can save you from having
to apply and undo a process multiple times just to find the right settings, making it a valuable
time-saving feature.
Another advantage to using the Compare Settings tool is seeing the effect your settings have in the
spectrogram/waveform display and spectrum analyzer.
 COMMON MODULE CONTROLS 66

• Process Comparison List: Each time settings are sent to the Compare window, a new list item
is created, by default titled “Settings 1,” “Settings 2,” etc.
• Preview: To hear (and see) the result of an item in the list, select that item and hit Preview.
• View Settings: Updates the controls to reflect the settings selected in the Process Comparison
List
• Remove: Remove an item from the list.
• Rename: Allows you to rename items with more descriptive names.
• Render: Apply the selected list item to the audio file.
 MODULE LIST 67

16. Module List

Overview
The Module List in the RX Audio Editor groups modules into the following categories: Repair,
Utility, and Measurement. The modules displayed in the module list can be customized by saving
a module list filter. The module categories can also be collapsed by clicking on the category
header.

Repair Utility Measurements

Ambience Match ADV Azimuth ADV Find Similar
Breath Control STD & ADV Dither STD & ADV Markers
Center Extract ADV EQ STD & ADV Spectrum
De-bleed STD & ADV EQ Match ADV Waveform Stats
De-click Fade
De-clip Gain
De-crackle STD & ADV Leveler ADV
De-ess STD & ADV Loudness Control STD & ADV
De-hum Mixing
De-plosive STD & ADV Normalize
De-reverb STD & ADV Phase
De-rustle ADV Plug-in
De-wind ADV Resample STD & ADV
Deconstruct ADV Signal Generator
Dialogue Contour ADV Time & Pitch STD & ADV
Dialogue De-reverb ADV Variable Pitch STD & ADV
Dialogue Isolate ADV Variable Time STD & ADV
Guitar De-noise STD & ADV
Interpolate STD & ADV
Mouth De-click STD & ADV
 MODULE LIST 68

Repair Utility Measurements

Music Rebalance STD & ADV
Spectral De-noise STD & ADV
Spectral Recovery ADV
Spectral Repair STD & ADV
Voice De-noise
Wow & Flutter ADV

Saving module list filters

1. From the menu on the right of the List Filter Selector, select “Add List Filter”
2. Choose a name for your new List Filter, hit return/enter to save the name
3. Use the checkboxes to select the features you want to include
4. Click the “Done Editing" button to save your new filter
 MODULE CHAIN 69

17. Module Chain

Overview
The Module Chain section allows for the processing of multiple modules in series. You can choose
from a variety of factory presets to get started or configure your own custom chain and save a
preset to use it again in the future. This section includes options for adding, removing, rearranging
and customizing the processing of a module.

Controls

Name Description
Preset Menu Save or load a module chain preset. More information about
saving presets is covered in the Common RX Module Controls
chapter.
Add Module Click the [+] to add a module from the module list to the end of
the module chain.
Remove Module Click the “X” icon to remove the associated module from the
module chain.
Power Button Enables/disables module processing.

Reorder Module Click and drag a module panel up or down within the Module
Chain to change its order in the signal flow.
View Module Opens the module window with settings loaded for the
associated step in the module chain.
Frequency Options for processing a module only over a selected frequency
Selection range.
Settings
 MODULE CHAIN 70

NOTE
Wow & Flutter is not available in the Module Chain.
 AMBIENCE MATCH 71

18. Ambience Match

ADV | Module and Plug-in (Audiosuite only)

Overview
The Ambience Match module lets you match the noise floor of one recording to another recording.
For example, you can apply the ambience of a location recording to your ADR tracks.

Controls

• AMBIENCE THRESHOLD: Discards dialogue or noises above the ambience threshold when
learning for complex ambience profiles. Complex mode only.
• AMBIENCE TYPE: Chooses between Complex or Static ambience.
• MOVEMENT: Higher values have more variety. Lower values remove distinctive elements for a
more uniform sound. The elements that are removed as the movement value is reduced are
greyed out in the Ambience Match spectrogram. Complex mode only.
• RANDOMNESS: Higher values allow the generated ambience to differ more from the learned
input and encourage more variation. Lower randomness will try to match the learned input
closely and may sound more like copy-pasting. The default Randomness setting will typically
provide the most natural sounding results.Complex mode only.
• GAIN: Trims the level of the generated ambience.
• OUTPUT AMBIENCE ONLY: When checked, the selected region will be replaced by the
generated ambience. When unchecked, the ambience will be mixed with the selected audio.
 AMBIENCE MATCH 72

Ambience Type

Complex
Ideal for variable ambiences with movement and texture, Complex mode generates a more
dynamic ambience based on your learned source. Each time you render, Complex mode will
generate a new random ambience.

Static
Perfect for unchanging ambience like room tone, air conditioners, or fans, the static Ambience
Match algorithm analyzes an audio selection and finds the lowest common denominator (the noise
that is common across the audio file) and treats that as the ambient profile.

Workflow

Using Ambience Match module in the RX Audio Editor

To train Ambience Match, provide it with a selection of raw noise. If there is no single fragment of
raw noise, or you want to save time, selections with speech can be used. In Complex mode, the
Ambience threshold can be used to ignore speech or other loud sounds in the selection. In Static
mode the algorithm will intelligently discard speech and only leave noisy parts in the noise print.

The Spectrogram in Ambience Match

Ambience Match has a helpful spectrogram that reflects your learned noise profile. In Static
mode, the spectrogram will display what will be rendered after you learn. In Complex mode,
the spectrogram will display your learned audio, and takes into account the ambience threshold
setting.
To match the ambience between selections:
1. Open the Ambience Match module in the module list
2. Make a selection in a file.
3. Click Learn.
4. Make another selection.
5. Adjust the Gain level as desired. The Gain control adjusts the level of synthesized ambience.
6. Select Output Ambience Only if you want the selection replaced with only the ambience from
the first selection.
7. Click Render.

NOTE
The Ambience Match module cannot reduce the amount of ambience that already
exists in the selection, it can only increase it. To reduce the ambience, use the Spectral
De-noise module.

To create an Ambience Match preset:

 AMBIENCE MATCH 73

1. In Ambience Match, click the gear icon to the right of the preset drop-down menu.
2. Select Add Preset.
3. Enter the name for the new preset.
4. Press Enter.

Using Ambience Match as an AudioSuite Plug-In

In addition to applying Ambience Match inside of the RX Audio Editor, it can also be used as an
AudioSuite plug-in inside of Avid’s Pro Tools or Media Composer.
When using Ambience Match inside of Pro Tools or Media Composer, we recommend not learning
from audio that contains fades within the selection or the handles. As Ambience Match establishes
an ambient profile using the lowest common denominator, learning from audio that’s being faded
in may result in inconsistent detection of the noise floor. Handles can be preserved by using
Ambience Match in clip-by-clip mode.

NOTE ABOUT LEARNING AMBIENCE ON FADES

When using Pro Tools, you may see an inconsistent result as a result of Pro Tools
adding dithering to fades, which varies based on the session’s bit depth. Since this
dithering noise almost certainly doesn’t match the material’s noise profile, this will
throw off Ambience Match. If you’re running a session in 16 bits, the dither added by
a fade will be sufficient enough to affect the detection algorithm. The problem is less
pronounced in 24 or 32 bits. To adjust the bit depth of your session, go to Setup >
Session in Pro Tools.
 BREATH CONTROL 74

19. Breath Control

STD & ADV | Module & Plug-in

Overview
The Breath Control module intelligently detects breaths in dialogue or vocal recordings and
suppresses them. Removing and reducing breaths in recordings can be a time-consuming process
for dialogue editors and music producers alike. Breath Control can help reduce the time spent on
repetitive editing without sacrificing the quality of your dialogue or vocal recordings.
The Breath Control module analyzes your file and distinguishes breaths from dialogue or sung
vocals based on their harmonic structure. If any piece of the incoming audio matches a harmonic
profile similar to a breath, the module will suppress that portion of the audio until sung vocals are
detected. Different from a ‘Threshold’ based process in which the module is only engaged once the
audio has risen to a certain volume, Breath Control will perform its analysis regardless of level. This
allows for accurate breath recognition for a multitude of quiet or loud vocal styles with minimal
adjustment of the module’s controls.

Controls

• GAIN MODE: When Gain Mode is selected, the Breath Control module will reduce the gain of
detected breaths by an absolute amount, regardless of the level of the breath. In some cases,
this is desirable when trying to handle heavy breathing or as a way of removing all breaths from
a particular spoken or sung vocal take. Depending on your settings, this can result in unnatural
sounding results as the very quiet breaths may be inaudible, while the loud breaths will be
reduced to a normal level.
• GAIN (dB): Sets the desired amount of gain reduction applied to all detected breaths,
regardless of level.
 BREATH CONTROL 75

• TARGET MODE: When in Target mode, the reduction amount of the ‘Target’ slider will set the
level of detected breaths. This can result in more natural sounding breath reduction as detected
breaths are only reduced when necessary. Loud and abrasive breaths will be reduced heavily,
while quiet, natural sounding breaths will be left at the same volume.
• TARGET LEVEL [dBFS]: Sets the resulting desired level of all detected breaths above the set
target.
• SENSITIVITY: This controls how sensitive the breath control plug-in is when detecting the
harmonic structure of breaths in your audio.
• OUTPUT BREATHS ONLY: When enabled, only the audio of the detected breaths will be passed
to the output of the module. This can help when setting the Sensitivity control in order to make
sure that only the breathing in your audio is being processed.
 CENTER EXTRACT 76

20. Center Extract

ADV

Overview
The Center Extract module preserves (using Keep Center) or removes (using Keep Sides) the
center channel of a stereo file. Extracting the center will retain the center of a stereo field and
attenuate everything on the sides, such as signals panned to the left or right. See the Use Cases
section below for more contextual examples and additional information about Center Extract
processing.

Controls

• KEEP CENTER: When the signal you want to preserve is even in both channels and noise is
uneven between channels, extracting the center can remove a lot of noise.
• KEEP SIDES: If you want to preserve the wide stereo information and remove the center
information, you can keep the sides of the signal instead.
• ALGORITHM: Two different algorithms are available:
• TRUE PHASE: Cancels the center with phase information and retains the original
panning of the sides.
• PSEUDO PAN: Extracts the side information and artificially stereo-izes it into two
channels.
• REDUCTION STRENGTH: Controls the level of the preserved signal. Lower values will retain
more information, higher values will discard more information.
• ARTIFACT SMOOTHING: Helps to reduce or eliminate the “musical noise” that is often
characteristic of FFT-based processing. Musical noise can be described as how something
may sound underwater. Increase this slider if your output sounds watery, but decrease it when
too much smoothing makes your audio sound dull.
What is an FFT?
Fast Fourier Transform: a procedure for the calculation of a signal frequency spectrum. The
greater the FFT size, the greater the frequency resolution, i.e., notes and tonal events will be
clearer at larger sizes. However, when using FFT-based processing, the more audio you remove
from your source, the more likely you are to create undesirable artifacts.
 CENTER EXTRACT 77

• DRY MIX [%]: Controls the amount of unprocessed signal mixed into the processed signal.
Useful for reducing artifacts introduced by processing by preserving the original characterisitics
of your audio.

TIP: USE AZIMUTH BEFORE CENTER EXTRACT FOR BEST RESULTS

It is often a good idea to make sure stereo channels are balanced by running Azimuth
correction before using Center Extract.

Center Extract is not available when Composite View is active.

Center Extract processing is not available on mono files. The nature of Center
Extract processing makes it unapplicable to mono files because they lack stereo field
information.

Use Cases
• Using Center Extract as an alternative to Mid-Side Encoding: Center channel extraction will
preserve a stereo image if the side channels are retained. This can make it more desirable
in some cases than Mid-Side encoding (which would sum left and right hard pans into one
channel).
• Use Keep Center to Reduce noise in stereo files transferred from a mono source: A mono
record transferred to a stereo tape would have side channel noise that would be suppressed by
extracting the center channel using Keep Center.
• *Use Keep Sides to Remove vocals from a stereo recording:
• The lead vocal track, in many popular mixes, is typically panned to the center of the mix.
Panning something to the center results in equal information being present in the side (Left
and Right) channels.
• Using the “Keep Sides” processing mode will retain the unique side channel information
present in a file, and reduce the Center channel.
• This is useful for karaoke-style removal of vocals from a song, especially because the
process results in a coherent stereo image.
 DE-BLEED 78

21. De-bleed
Overview
De-bleed reduces the leakage of one signal into another, such as when vocals bleed into a guitar
microphone, or when a click track fed into headphones bleeds into an open mic. The De-bleed
module learns a bleed relationship between two tracks. In the descriptions below, these two tracks
are referred to as:
• BLEED SOURCE TRACK: This track contains only the bleed source audio.
• ACTIVE TRACK: This track contains the audio present in your source track (bleed), mixed in
with the audio you want to preserve.
De-bleed relies on a relationship existing between your Active track and Bleed Source track in order
to function properly and provide the best results. In order to properly establish the relationship
between the Active and Bleed Source tracks, the following requirements must be met:
• At least two files are open in the RX Audio Editor: De-bleed requires 2 tracks (a Bleed Source
Track & an Active Track) to process.
• The Bleed Source and Active Track have matching sample rates: The sample rates of the
Bleed Source Track and Active Track must match in order to process using De-bleed. Using
tracks of different sample rates may indicate that the tracks are not related.

TIP: HOW TO CORRECT SAMPLE RATE DIFFERENCES

If you need to correct the sample rates of your bleed source and active track
selections in the De-bleed module, you can use the Resample module in the RX
Audio Editor to correct sample rate differences between your files, if necessary.

• The Bleed Source and Active tracks are time aligned:

• The Bleed Source Track and Active Track must be time aligned within a few milliseconds of
each other.
• Time aligned means that if the two files were played back together, they would sound in
sync. So if the same audio events are occurring at the same points in the timeline for both
the Bleed Source and Active tracks, this means they’re time aligned.
You can adjust the length or timing of files in the RX Audio Editor in the following ways:
▶ To remove the active selection, use the Cut operation from the Edit menu. The key commands
are Command+X (Mac) or Ctrl+X (Windows).
▶ To adjust the length of your file, apply Insert Silence using the Signal Generator module.
 DE-BLEED 79

Controls

• BLEED SOURCE TRACK SELECTION MENU: Select the Bleed Source track from this dropdown.
• ACTIVE TRACK: Displays the name of the track tab you are currently viewing.
• LEARN: Learns the relationship between the two tracks.
• BLEED PROFILE DISPLAY: After learning the bleed profile, this displays a portion of the
relationship captured between the two tracks. Toggle between the Bleed Source and Active
track displays to check that the bleed present in the bleed source track is present and aligned
with the bleed in your active track.
Bleed source display:

Bleed active track display:

 DE-BLEED 80

• STRENGTH: Determines the amount of bleed reduction applied during processing. Higher
Strength values may result in the removal of audio you wish to preserve. It is recommended you
start with lower strength values and increase the values if needed to achieve the most desirable
results.
• ARTIFACT SMOOTHING: Helps to reduce or eliminate the “musical noise” that is often
characteristic of FFT-based processing. Musical noise can be described as how something
may sound underwater. Increase this slider if your output sounds watery, but decrease it when
too much smoothing makes your audio sound dull.

WHAT IS AN FFT?
Fast Fourier Transform: a procedure for the calculation of a signal frequency
spectrum. The greater the FFT size, the greater the frequency resolution, i.e., notes
and tonal events will be clearer at larger sizes. However, when using FFT-based
processing, the more audio you remove from your source, the more likely you are
to create undesirable artifacts.

Workflow
1. Import the bleed source file (Bleed Source Track) and the file you are removing bleed from
(Active Track) into the RX Audio Editor. Ensure that the files are in sync (instructions above).
2. Open the De-bleed module.
3. Ensure that the Active Track name that is displayed in the De-bleed module is the file you wish
to modify.
4. In the De-bleed module, select the Source Track from the Bleed Source Track drop-down
menu.
5. In the active file tab (Active Track), make a selection where the bleed is most obvious.
6. Click the Learn button in the De-bleed module. Learn analyzes the general relationship between
the Bleed Source and the Active Track.
7. After the Learn pass is complete, select part or all of the Active Track and click Process.
 DE-CLICK 81

22. De-click
Module & Plug-in

Overview
The De-click module’s sophisticated algorithm analyzes audio for amplitude irregularities and
smoothes them out. This means that you can use De-click to remove a variety of short impulse
noises, such as clicks caused by digital errors, mouth noises, and interference from cell phones.

Controls

• ALGORITHM: Provides options for adjusting the configuration and processing quality of click
interpolation, depending on the types of clicks and pops present in the audio.
• SINGLE BAND: Processes quickly and works well on very narrow “digital” clicks
• MULTI-BAND (PERIODIC CLICKS): Uses multi-band processing for removing regularly
repeating clicks with a wider spectrum, or regular clicks that have concentrated low or
high energy (like thumps or optical soundtrack perforation noise)
• MULTI-BAND (RANDOM CLICKS): Uses multi-band processing for wider vinyl clicks and
thumps, with a protective algorithm for preserving periodic audio elements characteristic to
certain instruments such as brass or vocals
• LOW LATENCY: Works well on mouth clicks and other clicks that cannot be handled by
other algorithms. This mode has very low latency and is suitable for real-time work in RX
De-click plug-in.
• SENSITIVITY: Determines how many clicks are detected in the signal. Increasing sensitivity
can impact plosives, which can in turn reduce or damage the original signal.
• FREQUENCY SKEW: Adjusts the weighting of click removal toward higher or lower frequency
clicks. Negative values are more suitable for generic clicks such as those found on vinyl
recordings. A setting of zero or above targets mouth clicks in the middle frequencies.

NOTE
Frequency skew is not available when using the Single Band Algorithm option.

• CLICK WIDENING: Extends the repair area around detected clicks, compensating for mouth
sounds such as lip smacks that have a decay.
• OUTPUT CLICKS ONLY: Outputs the difference between the original and the processed signals
(suppressed clicks).
 DE-CLICK 82

Instant Process Tool

The Instant Process Tool offers a smart De-click mode, which instantly applies the active settings
from the De-click or Interpolate modules. Simply put, you may make any selection, and this mode
will automatically remove all clicks present in that selection. This is particularly useful for editing a
dialogue file, mismatching sample rate clicks and pops, and vinyl clicks.
If you’ve made a selection under 4000 samples in length, Instant Process will automatically use
the Interpolate algorithm. Interpolate will fill the selection with audio information based on the
surrounding audio.
For selections above 4000 samples, Instant Process will use the current settings of the De-click
module. De-click is effective on selections above 4000 samples in size, identifying clicks in relation
to desirable audio, and intelligently separating and removing the clicks.
For example, if the De-click module is indicating that a preset named ‘Remove mouth clicks’ is
loaded, these settings will be applied every time you use the Instant Process Tool in ‘De-click’ mode
(on selections longer than 4000 samples).
 DE-CLIP 83

23. De-clip
Module & Plug-in

Overview
De-clip repairs digital and analog clipping artifacts that result when A/D converters are pushed too
hard or magnetic tape is over-saturated. It can be extremely useful for rescuing recordings that
were made in a single pass, such as live concerts or interviews, momentary clipping in “perfect
takes”, and any other audio that cannot be re-recorded.
De-clip will process any audio above a given threshold, interpolating the waveform to be more
round. Generally, the process is as easy as finding the clipping you want to repair, then setting the
threshold just under the level where the signal clips.

Controls

• HISTOGRAM METER: Displays waveform levels for the current selection as a histogram. A
histogram is an analytical tool that displays how many samples are present at a given signal
level over a window in time. The longer the line for the histogram is, the more energy is
present at that amplitude. If a lot of energy tends to collect near the top and bottom edges
of a waveform, that waveform is probably clipped and distorted. The histogram meter helps
you set the Threshold control by displaying the audio level where the waveform’s peaks are
concentrated. This usually indicates at what level clipping is present in the file.
 DE-CLIP 84

• HISTOGRAM ZOOM CONTROLS: The histogram’s range can be scaled if you need a better view
of your signal. Use the (+) and (-) buttons to scale your display and value resolution for the
De-clip module. These buttons reduce (+) and/or expand (-) the range of the threshold slider
and histogram. You may want to extend the histogram range when the clipping point is lower
than what you can see on a histogram or if you don’t see anything on the histogram.
• In the RX Audio Editor, the Histogram meter updates based on selection: Select a section of
the recording where clipping is prominent and De-clip will analyze the levels of the program
material. If clipping is present in the selection, it will usually appear as a horizontal line in the
histogram that extends all the way across the meter.
• In the De-clip Plug-in, the histogram runs as a real-time meter.
• THRESHOLD [dB]: Defines the level used for detection of clipped intervals. Generally, this
should be set just below the actual level of clipping. To set the threshold, move the Threshold
slider until it lines up with the place in the histogram just below where clipping is concentrated.
Adjusting the Clipping Threshold will display a blue line within the histogram and a gray line on
the waveform itself (when the De-clip Threshold effect overlay). These lines indicate the audio
information that will be considered as “clipping” by the de-clip algorithm.
• THRESHOLD LINK: Toggles the ability to adjust positive and negative clipping thresholds
independently.

• When this option is enabled, you can adjust the positive and negative clipping Threshold
controls independently. This is useful in cases where more clipping is occurring on one side
of your waveform.
• You can also set asymmetric de-clipping thresholds directly from the waveform by toggling
the lock box between the threshold controls on the waveform display.
• SUGGEST: Calculates suggested threshold values based on the levels in your current selection.
• QUALITY: Controls the interpolation processing quality. There are three quality modes in the
De-clip module: Low, Medium, and High.
• When enabled and the De-clip module is open, the De-clip threshold overly will be displayed
in the spectrogram/waveform display.
• High quality mode processes slowly but is capable of achieving better results.
• In many cases you will find that Low quality mode gives you great results. To save time,
always start by previewing the Low quality modes first. You can also use the Compare
feature to try multiple modes and preview the results.
• MAKEUP GAIN [dB]: Selects the gain to be applied to the selection after De-clip. Since the
De-clip process causes an increase in peak levels, the Makeup gain control can be used to
prevent the signal from clipping after processing. It is also useful for matching the level after
processing to unprocessed audio outside of the selection.
• POST-LIMITER: Applies a true peak limiter after processing to prevent the processed signal
from exceeding 0 dBFS.
• De-clip usually increases signal levels by interpolating signal segments “above” the clipping
point, which can make the signal clip again if the waveform format offers no headroom
above 0 dBFS.
• If the post-limiter is disabled, the restored intervals above 0 dBFS can be safely stored even
without makeup gain as long as the file is saved as 32-bit float. Intervals above 0 dBFS will
clip when played back through a digital-analog converter.
 DE-CLIP 85

Using the De-clip Threshold Effect overlay in the Spectrogram/waveform

view
• By default, De-clip Threshold is enabled in the View > Effect Overlay menu.
• When enabled and the De-clip module is open, the De-clip threshold overly will be displayed in
the spectrogram/waveform display.
• You can adjust the Threshold controls in the De-clip module by adjusting the Threshold overlay
lines in the spectrogram display.
• You can use the mousewheel on the waveform amplitude ruler to adjust the threshold control
values.

More Information

Suggestions for severe distortion

• For certain situations, using the Deconstruct module to extract the noise components of the
distortion can help remove additional artifacts beyond the clipped peaks in a waveform.
• In cases where severe distortion is visible on the spectrogram, the Spectral Repair tool can be
used to select those problem areas, and attenuate or replace them with undistorted audio.

Visual Examples

Before & After Clip Repair

A waveform before and after clip repair. The after example (bottom) shows the original repaired
waveform (faded) and the post-limiting waveform (bright).

Unlinking Threshold Controls To Curb Asymmetric Clipping

This problematic waveform (grey) shears off around −13 dB on only the positive side of the
waveform (the histogram on the right shows the extra positive energy of clipped audio). Extra
processing of the negative side would be unnecessary, so the Threshold controls can be unlinked
to process above −13 dBFS on the positive side only. The resulting waveform is drawn in blue
above the grey sheared peak.
 DE-CRACKLE 86

24. De-crackle
Module & Plug-in

Overview
When an audio signal contains many clicks that are close together and low in volume, this is
described as crackle. De-crackle is very effective at removing these audio problems and even more
effective if it is run after De-click has eliminated the worst offending clicks.

Controls

• QUALITY:
• LOW quality offers fast processing
• MEDIUM quality will remove periodic, rapidly repeating clicks
• HIGH quality will help preserve the tonal qualities of a signal
• STRENGTH: Controls the amount of crackle that is detected and repaired.
• AMPLITUDE SKEW: Biases the processing toward higher or lower volume crackle. If the crackle
accompanies transients and other high-level signal passages (such as during clipping), set this
control more to the right. If the crackle mostly happens during low-amplitude signal passages,
set this control more to the left.
• OUTPUT CRACKLE ONLY: Outputs the difference between the original and processed signals
(suppressed crackle).
 DE-ESS 87

25. De-ess
Module & Plug-in

Overview
De-ess attenuates or reduces sibilance, the harsh high-frequency sounds that come from “S,” “F,”
“X,” “SH,” and a soft “C.”

Controls

• MODES: The De-ess module offers two processing algorithms:

• CLASSIC MODE: Detects sibilants and attenuates them with a broadband gain envelope.
Since attenuation is applied to all frequencies, this mode is less targeted than Spectral
De-ess.
• SPECTRAL MODE: Offers a more transparent, intelligent, and frequency-specific type of
de-essing than Classic Mode. Spectral Mode only attenuates the high frequencies where
sibilance is most active, leaving the lower frequencies untouched. Technically, Spectral
De-ess is a multiband compressor with dozens of bands. It’s able to compress the level
of sibilants, shape their spectra, and avoid modulation of ambient noise. Each band can
operate independently or with a link to adjacent bands (adjusted by the Spectral Shaping
slider, explained below) and band thresholds can be adjusted for the desired shape of a
sibilant (achieved by adjusting the Spectral Tilt slider).
 DE-ESS 88

• THRESHOLD: Determines the level at which the De-ess module begins compressing sibilance.
The Threshold control has two modes that determine how it reacts to incoming signal level. It is
specified in decibels, relative to speech level (Relative Mode) or full scale (Absolute Mode).
• RELATIVE MODE: Determines the level of speech and sets the threshold relatively to that
level. This is the default Threshold mode, Relative mode is active when the “Absolute”
checkbox below the Threshold slider is not checked.
• ABSOLUTE MODE: Sets the threshold to a decibel level below full scale (dBFS). This mode
is enabled by checking the “Absolute” checkbox below the threshold slider.
• CUTOFF FREQUENCY: Specifies the crossover point between speech (to be preserved) and
sibilance (to be reduced). The Cutoff frequency value functions as the lower boundary for
sibilance detection.
• SPECTRAL FLATTENING: Spectral Flattening determines how much the spectral shape of the
sibilant is changed. Think of Spectral Flattening as a way to fine tune the strength of the sibilant
processing. The flatter you go, the more the sibilant is reduced. A setting of 0% leaves the
natural shape of the sibilance by applying uniform compression across all bands. A setting of
100% flattens the shape of the sibilant toward a specified noise profile (see Spectral Tilt).
• SPECTRAL TILT: Spectral Tilt creates a target noise profile for the sibilance. It gives you the
flexibility to determine what the ideal shape of your high frequency signal could be. Moving the
shape toward brown noise moves it toward a darker sound. Moving the shape toward white
noise provides a brighter result. Together with flattening and threshold, you can determine how
much shaping takes place. A setting of 0 creates a natural spectral decay similar to pink noise.
Values below or above 0 create a profile that is heavier in low frequencies (like brown noise)
or high frequencies (like white noise). Spectral Tilt is most effective when the Spectral Shaping
control is set to a non-zero value.
• SPEED: Sets the attack and release times for the processing. Attack times are program-
dependent in both modes.
• FAST: Uses quicker attack and release times. If De-ess is causing pumping in the high
frequencies, try using Fast mode to mitigate this problem. If the settings are too slow, the
processor won’t recover quickly enough, resulting in too much high frequency attentuation.
• SLOW: Uses longer attack and release times. If De-ess is smoothing out the transients
too much, try using Slow mode to mitigate this problem. Settings that are too fast can
reduce the high frequency signal too much in the initial transient phase and can introduce a
smoothing effect that reduces useful high frequency definition.
 DE-HUM 89

26. De-hum
Module & Plug-in

Overview
De-hum is designed to remove persistent tonal noise, like AC noise, electrical buzz or interference
introduced by EMF (Electro-magnetic fields). De-hum includes 2 modes: a Static mode that’s best
for ground noise or simple hums with a few harmonics, and a Dynamic mode for complex hums or
buzz.

Mode Selection
• Static:
• Ideal for simple tonal noise with only a few harmonics of the fundamental.
• Hum that doesn’t overlap the speech frequencies.
• Dynamic:
• Hum that has more than a dozen harmonics.
• Hum that is inharmonic (contains multiple unrelated tones).
• Buzz or EMF that extend into the upper frequencies.
• Hum/buzz that overlaps speech frequencies.
• Less Ringing due to gated notching

MORE INFORMATION ABOUT DYNAMIC MODE

For more information about the benefits of the Dynamic Mode in De-hum, read this
article by iZotope Principal DSP Engineer Alexey Lukin.
 DE-HUM 90

Controls in Dynamic Mode

• LEARN: Captures the full tonal profile for removal of complex hum. To learn in the app, make a
selection containing the hum in isolation and click the Learn button. In the plug-in, enable learn
and play back the section of audio containing the hum. If you cannot find a selection of hum in
isolation, RX can analyze any audio with prominent hum, but the results may not be as useful.
• ADAPTIVE MODE: Allows De-hum to adjust the notch filters based on changes in the audio
over time. In this mode, RX will analyze incoming audio to determine what is hum and what
is desired audio material. Adaptive mode will work better with hum that changes in pitch
throughout the file.

ADAPTIVE DYNAMIC MODE NOTE

Adaptive Dynamic mode in the De-Hum plug-in is not optimized for real time use.
In the plug-in, please use this feature offline.

• SENSITIVITY: Adjusts the amount of hum that will be removed.

• BANDS: Controls the number of dynamic notch filters, which update their gains based on the
input signal to avoid ringing. Adjust to higher values for complex hums, at the expense of
transparency.
• FILTER Q: Controls the bandwidth of the notch filters.
• RANGE SELECTORS: Sets the action region of the hum removal by defining the upper and
lower bounds of De-hum processing.
 DE-HUM 91

Controls in Static Mode

• BASE FREQUENCY: Sets the fundamental frequency of the hum to be removed. With Preview
engaged, adjust the slider until you find the point where the hum is appropriately reduced.
• FREQUENCY (Hz): Sets the fundamental frequency for the filter.
• FILTER Q: Controls the bandwidth of the notch filters for the base frequency and all of the
harmonics.

During Preview, you can use the Spectrum Analyzer to help identify the base
frequency of the hum.

• LEARN: Automatically set the Base Frequency based on the learned profile. To learn in the
app, make a selection containing the hum in isolation, and click the Learn button. In the
plug-in, enable learn and playback the section of audio containing the hum. If you cannot find
a selection of hum in isolation, RX can analyze any audio with prominent hum, but the results
may not be as useful.
• ADAPTIVE MODE: Allows De-hum to adjust the notch filters based on changes in the audio
over time. In this mode, RX will analyze incoming audio to determine what is hum and what
is desired audio material. Adaptive mode will work better with hum that changes in pitch
throughout the file.
 DE-HUM 92

• LINEAR-PHASE FILTERS: Linear-phase enables FIR (Finite Impulse Response) filters with a
high FFT (Fast Fourier Transform) size. These filters provide very accurate frequency response
with no change in phase at the expense of latency and filter pre-ringing.

DISABLING LINEAR PHASE (FIR) FILTERS

When Linear Phase is disabled, De-hum will use minimum-phase IIR (Infinite
Impulse Response) filters. These are also very accurate, and are only susceptible
to post-ringing, which is usually less noticeable than the pre-ringing introduced by
FIR filters.
Latency Consideration: Disabling Linear Phase Filters will reduce the latency used
by De-hum when it is being used as a real-time plug-in.

• HIGH/LOW-PASS FILTERS: These traditional filters come ahead of the De-hum notch filters,
and allow for frequencies to pass above or below a certain cutoff point. These can be useful for
tackling extreme hum or buzz.
• FREQUENCY (Hz): sets the cutoff frequency for the filter
• FILTER Q: Sets the bandwidth of the filter (or dB/octave cut). In the default IIR filter mode
with a high Q setting, you may notice a resonance at the cutoff frequency characteristic
of traditional analog filters. That resonance can be mitigated by engaging the Linear-phase
filters.
• HARMONICS: Harmonics often accompany the fundamental frequency of a hum. With De-hum
you can also attenuate these overtones with additional notch filters. Using the Harmonics
control, you can add up to 16 harmonics. The spectrogram display makes it easy to identify
the harmonics. After setting the number of harmonics, use the Slope control to set how
aggressively the higher harmonics are attenuated.
• LINK HARMONICS: Connects the gain controls of the notch filters.
• ALL: presents a single node on the display for controlling the gain of all the notch filters.
This is the default setting.
• ODD/EVEN: presents two nodes on the display, one for controlling the gain of the
fundamental frequency and even harmonics, and another for controlling the 1st harmonic
and any following odd harmonics.
• NONE: presents individual gain nodes for the fundamental and each harmonic.
• SLOPE: When harmonics are linked, this controls the harmonic slope of the gain nodes for each
overtone. As the harmonic order increases, the gain level resolves closer to 0 dB. When the
Link Harmonics control is set to Odd/Even, a separate control appears that affords independent
control over the slope for both odd and even harmonics.
• HARMONIC PANEL [dB]:
 DE-HUM 93

• ENABLE: This control allows you to choose which notch filters are active. Disabling
unnecessary notches can help preserve the original sound.
• GAIN [dB]: This controls the level for each notch filter. You can also manually enter gain
settings for the fundamental, or any of the harmonics if Link Harmonics is set to None.
• FILTER DC OFFSET: This checkbox will engage a filter to remove any DC (direct current) offset
that sometimes occurs in A/D converters or analog circuits used in the recording process.
• OUTPUT HUM ONLY: Selecting this check box will isolate the hum that is being removed. This
is useful for fine-tuning your settings. Identify a section of your file where the hum is mixed
with other material, select this mode, and click Preview. Now adjust parameters like Filter Q and
Slope control to maximize hum removal, thus minimizing the effect on the program material.

More Information

Ringing
Every notch filter can be characterized by its frequency response and its impulse response.
The frequency response determines how the filter is changing amplitudes of different frequency
components in the signal. The impulse response determines what the filter does to the signal
waveform and how it handles transients.
Increasing the Q of a notch filter makes it narrower, but adds more “tails” to the impulse response
of the filter. These tails are called ringing because they add a reverberant, ringing character to
the perfect click of the impulse. This ringing matches in frequency with the bends in the filter’s
frequency response.
Minimum Phase EQ approximates the behavior and sound of analog EQs and places all ringing
behind (after) the signal transient, where it has a higher chance of being psychoacoustically
masked by the signal itself. Since narrow notching in De-hum is likely to create significant ringing,
minimum-phase filters are usually the preferred type when addressing ringing issues after notch
filtering with De-hum.
 DE-HUM 94

Alternative Modules to use for Complex Hum Issues

The Dynamic mode of De-hum is a powerful tool for attenuating tonal noises prior to other
denoising algorithms that target random components of the noise: Voice De-noise or Spectral
De-noise. There are cases, however, when other modules may successfully replace or complement
De-hum.
When there is no time to run De-hum and then follow up with De-noise (for hiss), just using
RX Spectral De-noise can also take care of the hum rather efficiently. Spectral De-noise has
separate controls for reduction of tonal and broadband components of the noise. A manually
drawn reduction curve provides an additional way to fine-tune the amount of attenuation.
A quick fix for an occasional single-tone buzz is provided by Spectral Repair: just select the
offending frequency on a spectrogram and apply the Attenuate Vertically mode with a reasonably
high number of Bands (1024–4096).
Some other alternatives to De-hum:
• Guitar De-noise For guitar tracks containing buzz, try using Guitar De-noise. This module can
help reduce amp buzz along with string squeaks and picking that is too aggressive.
• Spectral De-noise For hum that has many harmonics that extend into higher frequencies
(often described as “buzz”), try using Spectral De-noise. Spectral De-noise features tonal noise
reduction controls that can make short work of harmonic hum and buzz across the entire
spectrum.
• De-click When the buzz is very harmonically rich, sometimes using De-click in either Single-
band or Multiband (periodic clicks) mode can treat the clicks that comprise buzz.

MORE INFORMATION ABOUT DYNAMIC MODE

For further information about RX De-hum and when to use it, refer to this article
authored by iZotope Principal DSP Engineer Alexey Lukin.

Visual Example
This image shows the spectrogram of a file with 3 harmonics of a 60 Hz Hum:
 DE-PLOSIVE 95

27. De-plosive
Module & Plug-in

Overview
A plosive is a consonant speech sound that occurs when the vocal tract briefly blocks the flow
of air during speech. When the block in the vocal tract is resolved, an audible release of pressure
occurs, called a plosive. Pop filters (in a studio) or wind filters (on location) are commonly used
when recording vocals or dialogue, in order to reduce the popping effect that can occur when the
higher pressure plosive signal comes into contact with the diaphragm of a microphone.
De-plosive can intelligently identify, separate and reduce plosives in an input signal while still
preserving the fundamental frequency content and harmonics of the dialogue.

Controls
The following section describes the controls available for refining plosive reduction.

RECOMMENDATION: USE DE-PLOSIVE BEFORE APPLYING A HIGH PASS FILTER

De-plosive attempts to detect plosives between 20 to 80 Hz. If the input file has
already been filtered with a high-pass filter, it is possible that plosive detection will not
function correctly. To achieve the best results, it is recommended that De-plosive be
used before applying a high-pass filter.

Sensitivity
Adjusts how much of the input audio will be categorized as a plosive. Sensitivity has a greater
effect on the overall effectiveness of plosive reduction than simply increasing the Strength control.
• Lower values will instruct the algorithm to narrowly define what it will detect as a plosive. This
may result in less overall reduction of plosives with the benefit of maintaining the quality of the
speech signal.
• Higher values will instruct the algorithm to more broadly detect plosives in the input signal,
which may result in too much of the speech signal being identified as, and subsequently
reduced with, the plosive signal content.
 DE-PLOSIVE 96

Strength
Adjusts the amount of reduction applied to the detected plosive signal content. Higher values can
result in significant plosive reduction at the cost of decreasing the quality and clarity of the speech
signal. To maintain the speech signal while still reducing plosives, try increasing the Sensitivity
control and decreasing the Strength control.

Frequency Limit [Hz]

Sets the upper frequency boundary for plosive reduction. Plosives typically manifest between 20Hz
to 300Hz, but may sometimes reach as high as 500Hz. Setting this control just above where the
plosives are occurring in the input signal can help reduce unwanted reduction of frequency above
where the plosives are occurring.

Identifying plosives in the spectrogram display

The spectrogram can be used to help determine the highest plosive frequency in the input signal.
Plosives are caused by small bursts of air pressure hitting the diaphragm of a microphone which
results in an increase in amplitude. The colors used to represent signals in the spectrogram are
tied to the amplitude of a signal. Darker colors represent lower amplitudes and brighter, more
vibrant colors represent higher amplitudes. Plosives will generally appear brighter than the signal
content that surrounds them.
 DE-REVERB 97

28. De-reverb
Module & Plug-in

Overview
De-reverb gives you control over the amount of ambient space captured in a recording. It can make
large cathedrals sound like small halls and make roomy vocals sound like they were recorded in a
treated space. De-reverb processes audio according to the reverberant/direct ratio (also known as
wet/dry ratio) detected in the signal. It can learn your audio to suggest some frequency and decay
time settings, or you can estimate these yourself.

Controls
 DE-REVERB 98

• LEARN: Teaches De-reverb how much reverb is in your signal.

• The Learn feature analyzes the signal and determines the wet/dry ratio per frequency of
your signal, as well as the overall rate of decay of reverb.
• When the Learn operation completes, the Reverb Profile and Tail Length controls will be set
to their suggested values.
• The Learn operation can be performed on any reverberant audio.

RECOMMENDATION
We strongly recommend using the Learn feature on a selection of audio that
starts with some noise floor (or room tone), is several seconds long, and
includes both the direct signal and its reverberant tails.

• METERING
• The top meter shows a comparison between the input and output signal energy over the
past five seconds of playback.
• The bottom meter shows the amount of reverb reduction over time. It is the difference
between input and output plotted on a flat line.
• Both of the meters together give you an idea of what De-reverb considers reverb and help
you refine your settings.
• REDUCTION: Controls the applied amount of De-reverb.
• Larger amounts mean more reverb is removed.
• Smaller amounts perform less processing.
• This control represents the target wet/dry ratio for processing. In other words, if it is set very
high, it will treat the signal as though it has more reverb and process it more.

NOTE: NEGATIVE REDUCTION VALUES

Negative Reduction values will increase the amount of reverb in the signal.

• REVERB PROFILE: Controls the amount of De-reverb effect applied per band.
• These controls are set automatically by the Learn feature.
• If a group of reverberant tones are more prominent in a signal, increase the control for that
band.
• Generally you want to set these controls to match the reverb originally present in your
signal. For example, if reverb takes longer to decay (or is more present) in a particular band,
set that control higher.
• These controls can also be used to address more prominent ringing or resonant groups
in signals. For example, increasing the profile control for low frequencies can remove
muddiness from a resonant bass guitar, while increasing the high band control can curtail
ringing sibilance in live vocal recordings.
 DE-REVERB 99

• TAIL LENGTH: Controls the decay of De-reverb processing. This control is an approximation of
RT-60, the rate of time it takes for a reverberant signal to decrease in amplitude by 60 dB. This
is automatically set by the Learn feature.
• Increase this control if reverb tails reappear after processing, or if early reflections are too
apparent.
• Decrease this control if reverb tails and noise floors sound over-processed, or if the
processed audio sounds dull.
• Setting this control to the minimum value is effective at processing early reflections.
• ARTIFACT SMOOTHING: Controls the frequency accuracy of De-reverb processing.

ARTIFACT SMOOTHING DEFAULT VALUE NOTE

Because reverb is generally smooth across the frequency spectrum, the default
value of this control is very high. However, if you need more accuracy to address
issues like resonant tones in a room, you can decrease this control. The tradeoff
is generally more artifacts from strong processing, so you may have to balance
adjusting this and the Reduction control.

• ENHANCE DRY SIGNAL: Increases the level of the direct signal.

• Boosting the direct signal can help create more dynamic range in the signal, and is a good
option to try when working with voice or transient material.
• Enabling this option can also help prepare material for later de-noising.
• OUTPUT REVERB ONLY: Changes the output of De-reverb from the processed signal to the wet
reverberant signal.
• This is useful for monitoring the processing to get better results. Hearing just the reverb
helps you understand the impact of controls like Reduction, Reverb Profile, Tail Length, and
Artifact Smoothing.
• When this option is enabled, the output may not sound much like reverb because it is the
difference of processing against the original signal, with some enhancement to expose
more of the reverb apparent in the recording.

More Information

Early Reflections
Early reflections are the rapid echoes of a direct sound from a nearby surface. They are often
distinct from the rest of a reverberant tail because they have a lot of energy but end quickly. Early
reflections typically comprise the first 5 to 100 milliseconds of a reverb tail.

Using De-reverb as a real-time plug-in

• De-reverb is available as a VST/AU/RTAS/AAX real-time plug-in.
• However, due to the complexity of this processing, it can be resource intensive.
• To achieve high-quality results, it is always best to bring the audio file in question into RX Audio
Editor (via RX Connect or by opening it directly), applying De-reverb, and then returning the file
back to your original session.
 DE-REVERB 100

Tips for Learning a Reverb Profile

• To find the best settings for your signal quickly, find about five seconds of audio that starts with
noise and has both direct signal and reverberant tails.
• If you can find enough direct signal and reverberant tails to fill the De-reverb signal trace
meter while using Learn, you will probably get a good reverb profile.
• Direct signal, reverberant tail, and noise are all important to help De-reverb understand your
audio and set its controls appropriately. It needs to understand the ratio of dry signal to
reverberant signal, how long reverb tails last, and where the noise floor of your signal is (to
avoid excessive processing).
• If you have trouble getting good results from the Learn feature, you can try:
• learning on transient broadband audio, like drums, claps, or coughs
• learning on any audio that is obviously reverberant
• learning for a longer amount of time. Most reverb can be analyzed in a few seconds, but
some reverb profiles can require up to ten seconds of analysis.
• Does De-reverb processing sound unnatural?
• If the output of De-reverb sounds unnatural after Learning, try slowly decreasing the
reduction control.

Visual Example of De-reverb processing

De-reverb has the effect of sharpening a signal in time. You can see this transition in the
spectrogram: reverberant audio looks blurred, and cleaned audio appears more focused. Here,
a recording of a distant speaker (left) has had its long tails processed (center) before another
De-reverb pass with shorter tail lengths to tackle the early reflections (right):
 DE-REVERB 101

Tips for dealing with complicated reverbs

If the audio you are working with has a very complex reverb, such as a reverb with apparent early
reflections, you may get better results after trying a few passes of De-reverb.
1. First, start by training the De-reverb and set the Reduction amount to a value that yields good
results on the long reverberant tail.
2. After processing, Learn a new reverb profile and try reducing the level of early reflections: set
the Tail Length control to 0.5, Artifact Smoothing around 3.0, and increase Reduction.
3. A combination of De-reverb and Spectral De-noise can be used to tame very reverberant
signals. It does not matter whether you process with De-reverb or Spectral De-noise first.
 DE-RUSTLE 102

29. De-rustle
ADV | Module & Plug-in (Audiosuite Only)

Overview
De-rustle is designed to reduce the noise or “rustle” that is often the result of a lavalier microphone
rubbing or brushing against clothing during a recording. This type of noise can vary unpredictably
over time and exhibit a wide variety of sonic characteristics from higher frequency “brushing” to
lower frequency “thuds,” making it a challenging issue to resolve.
The De-rustle module uses a machine learning algorithm trained on isolated rustle samples, clean
dialogue samples and dialogue samples with rustle. When processing, De-rustle leverages the
trained data to identify and separate the rustle from the dialogue.

Controls

Reduction Strength
Adjusts how sensitive the separation algorithm is to identifying dialogue in the input signal.
• Lower values allow for broader detection of dialogue in the input signal. This can result in
more rustle in the processed signal, but can increase clarity of dialogue in the processed output
signal.
• Higher values allow for stricter detection of dialogue in the input signal. This can result in
more significant reduction of rustle or lav noise, but can decrease the clarity of dialogue in the
processed output signal.

Ambience Preservation
Adjusts the amount of background noise to retain in the processed output. The separation
algorithm may include background noise and ambience in the signal it has detected as rustle.
This can result in unwanted reduction of unrelated background noise in the processed signal.
 DE-RUSTLE 103

• Lower values allow for broader detection and separation of rustle against other background
noise in the input signal. This can result in more background noise being included and reduced
with the separated rustle component.
• Higher values allow for stricter detection and separation of rustle against other background
noise in the input signal. This can result in less background noise being included and reduced
with the separated rustle component.

Separation Algorithm
The following separation algorithm modes are available in the De-rustle module.

Channel Independent
When this mode is selected, the separation algorithm is applied to the input audio channels
independently. Channel independent mode is the fastest Separation algorithm option. It offers
the most efficient real-time preview performance and processing speeds when working with the
De-rustle module in the RX Audio Editor.

Joint Channel
When this mode is selected, joint channel processing is applied to the input audio before
determining the separation between dialogue and rustle signal components. Joint Channel
mode offers higher quality separation results than Channel Independent mode, especially when
processing stereo files with similar content on both channels (correlated signals, strong stereo
image).

Advanced Joint Channel

When this mode is selected, joint channel and additional advanced processing is applied to
the input audio before determining separation between dialogue and rustle. Advanced Joint
Channel mode offers the highest quality separation results, especially when processing files with
high sampling rates. This mode requires longer processing times than the other two modes. If
processing time is a concern, Channel Independent mode can be used as a faster, lower quality
alternative.
 DE-WIND 104

30. De-wind
ADV

Overview
De-wind removes intermittent low frequency rumble that can occur when light to moderate bursts
of wind come into contact with a microphone diaphragm. De-wind is not designed to remove heavy
wind bursts that blow out and distort a microphone signal.

Controls

Reduction
Determines the balance between the depth of wind reduction and the preservation of the original
signal.

Crossover Frequency
Sets the upper frequency limit for the De-wind processing algorithm.

Fundamental Recovery
Re-synthesizes lower voice harmonics that may be lost or obscured by wind.

Artifact Smoothing
Eliminates “musical noise” that is often characteristic of FFT-based processing. Musical noise can
be described as how something may sound underwater. Increase this slider if your output sounds
watery, but decrease it when too much smoothing makes speech sound muffled.
 DE-WIND 105

WHAT IS AN FFT?
Fast Fourier Transform: a procedure for the calculation of a signal frequency
spectrum. The greater the FFT size, the greater the frequency resolution, i.e., notes
and tonal events will be clearer at larger sizes. However, when using FFT-based
processing, the more audio you remove from your source, the more likely you are
to create undesirable artifacts.

Tips

When to use the De-wind Module

The De-wind module is best suited to remove and reduce intermittent wind noise in the foreground
(for example, a recording with periodic wind gusts that come into direct contact with the
diaphragm of the microphone) as opposed to constant wind noise in the background.
The De-wind algorithm attempts to preserve the noise floor of a recording by tracking how it
changes over time. Specifically, De-wind is looking for noise floor changes that are characteristic of
the noise introduced when a burst of wind comes in contact with the diaphragm of a microphone.
If the wind in your recording is constant or in the background, De-wind will treat it as part of the
desirable noise floor and will preserve it rather than remove it.

Alternative Modules
For recordings with constant or background wind noise, you may achieve better results by:
• Using the Spectral De-noise module.
• Running De-wind processing more than once.
 DECONSTRUCT 106

31. Deconstruct
ADV

Overview
Deconstruct analyzes your audio selection and separates the signal into Tonal, Noisy, and
Transient (optionally) audio components. The separate components of the signal can then be
cut or boosted individually using their associated Gain control.

Controls

Tonal Gain
Adjusts the level of the tonal components of the signal. Boosting a tonal signal (voice or
instrumental) can help lift it out of a noise floor.

Noisy Gain
Adjusts the level of noisy components of the signal. This can be very useful for highlighting areas
of raspiness or distortion only, and then attenuating the noisy gain to reduce overall distortion.
 DECONSTRUCT 107

Separate Transients
Enables transient separation processing and activates the Transient Gain control.

Transient Gain
Adjusts the level of transient components of the signal. This can work as a transient shaper or a
declicker, allowing you to attenuate or boost clicks and attacks.

TRANSIENT SEPARATION PERFORMANCE

The Separate Transients option allows you to control the level of transients, but
increases CPU load. This may impact the performance of Preview. In this case, using
the Compare functionality instead of Preview is recommended.

Tonal/Noisy Balance
Modifies the default weighting of the separation algorithm used by Deconstruct to categorize
components of a signal as either “noisy” or “tonal.”
• Negative values (Tonal weighting) will classify more of the “noisy” components of a signal as
“tonal” components and apply Tonal Gain to them during processing.
• Positive values (Noisy weighting) will classify more of the “tonal” components of a signal as
“noisy” components and apply Noisy Gain to them during processing.

Artifact Smoothing
Reduces “musical noise” artifacts that are often characteristic of FFT-based processing. Fast
Fourier Transform (FFT) is a procedure for the calculation of a signal frequency spectrum. The
greater the FFT size, the greater the frequency resolution, i.e. notes and tonal events will be clearer
at larger sizes. However, when using FFT-based processing, the more audio you remove from your
source, the more likely you are to create undesirable artifacts. Increase this slider if Deconstruct’s
output sounds watery, but decrease it when too much smoothing reduces the separation between
signal components.

More Information
• Deconstruct can be useful for a variety of audio files and applications, particularly when
attempting to remove noise that varies throughout the length of a file.
• Deconstruct differs from the Spectral De-noise and Voice De-noise modules, which separate
signal from noise based purely on amplitude. Deconstruct analyzes the harmonic structure of a
signal independently of level. It does not matter if a tonal signal like hum is quiet or prominent.
Deconstruct will treat it as a tonal component and adjust its gain accordingly.
• Deconstruct can be effective in removing residual vinyl noise that may be present after applying
De-click or De-crackle processing. Using Deconstruct in this situation may produce better
results than using the Spectral De-noise or Voice De-noise modules.
 DIALOGUE CONTOUR 108

32. Dialogue Contour

ADV

Overview
Dialogue Contour allows for the manipulation of the pitch envelope of a dialogue selection. It
features pitch correction processing that is tailored to speech. It is useful for adjusting the
inflection and/or expressiveness of words that may not flow or fit with the rest of the dialogue
in a clip. It can also be used to make a series of dialogue edits sound more cohesive.

Displays
Dialogue Contour features a waveform panel and a spectrogram panel that each display
information about the current selection in the active file tab. These panels will dynamically update
when the selection is changed. If no selection is made in the active file tab, no information will be
displayed in the spectrogram or waveform panels.

WINDOW RESIZING
Click and drag on the bottom right-hand corner of the module window to customize
the window size.

Waveform and Pitch Curve Display

The single waveform drawn in this panel represents a sum of all enabled channels in the current
selection. The waveform drawing is normalized to allow for consistent vertical resolution when
working with selections of varying amplitude.
 DIALOGUE CONTOUR 109

Dialogue Contour now has two curves displayed over the waveform view. The grayed out curve
is the original pitch of the selected audio. The orange curve is the projected pitch based on
parameters set by the user and will update to reflect any subsequent changes.

Spectrogram Display
The spectrogram drawn in this panel represents a sum of all enabled channels in the current
selection.

Playhead Indicators
The solid white vertical line and dotted yellow vertical line overlaid on the waveform and
spectrogram panels indicate the current playhead position (white) and the playhead anchor
position (yellow).

Contour Curve
The white line overlaid on the spectrogram panel represents the pitch contour curve. Nodes can be
added to this curve and adjusted to make changes to pitch over the course of the active selection.

Contour Curve Axes

The contour curve allows for adjustments along two axes: Pitch and Time.
 DIALOGUE CONTOUR 110

• PITCH: The vertical y-axis of the contour curve represents pitch in semitones.
• The Pitch axis ranges from -6 (bottom) to +6 (top) semitones.
• The center of the Pitch axis equates to 0 semitones.
• TIME: The horizontal x-axis represents time.
• The time format used here is determined by the time format display selection in the
transport section of the main editor window.
• The range of the time ruler matches the length of the current selection.
To zoom the ruler in and out:
1. Hover over the ruler and use a mousewheel or trackpad to zoom in and out.
2. Click and drag left or right on the ruler when zoomed in to change the ruler position.
3. Double-click on a ruler display to reset the zoom level to default.

Contour Curve Readout

When the cursor is positioned over the spectrogram panel, a text readout will appear in the upper
left hand corner of the panel. This readout displays information about the processing that will be
applied when the contour curve is rendered.

The readout displays the following information about the cursor position, from left to right:
• TIME: Current time position of the cursor within the spectrogram panel.
• PITCH SHIFT (%): Percentage of pitch shift that will be applied at the cursor’s current time
position.
• PITCH SHIFT (Semitones): Amount of pitch shift that will be applied at the cursor’s current
time position.

Contour Curve Editing

The following section describes the methods and controls available for editing the contour curve.

Add Nodes
Click in the spectrogram panel to add a new node to the contour curve.

CONTOUR CURVE NODE LIMIT

The contour curve supports adding up to 25 nodes.

Semitone Adjustments
Click and drag a node up or down to adjust its semitone value.
 DIALOGUE CONTOUR 111

Time Adjustments
Click and drag a node left or right to move it earlier or later in time.
• Nodes cannot be moved outside of the time bounds of the current selection.
• The contour curve shape will be maintained when the selection changes.
• The contour curve shape will be maintained after rendering.

Remove Nodes
Individual nodes can be deleted from the curve using the following methods:
▶ Click and drag a node past the top or bottom edge of the contour curve display to quickly
remove it from the curve.
▶ Control-click (Mac) or right click (Windows) on a node to remove it from the curve.

Reset Individual Nodes

Double-click or Alt-click (Option-click) on a node to reset it to default (0 semitones).

Controls
The following section describes the controls available for refining formant scaling and applying a
global pitch offset to the entire selection.

Pitch
Applies a global semitone offset value to the current selection. This value is added to or subtracted
from the processing applied by the contour curve. Adjusting the Pitch amount will not update the
contour curve display.

The Pitch slider can be useful for adjusting a single word by a static amount. For
example, setting Pitch to +2 semitones with the contour curve set to default will shift
the pitch of the current selection up by 2 semitones.

Formant
Applies a global formant shift (in semitones) to the current selection. Formants are resonances of
a vocal tract that define the timbre of a voice. They are independent of pitch, so adjusting formants
will not update the pitch curve display.
Shifting formants up makes the voice sound thinner, while shifting them down makes the voice
sound heavier and bigger.

Variation
Changes the amount of expressivity in a voice by scaling the existing pitch variations. When the
Variation is set to -100%, the pitch becomes flat (monotone). 0% corresponds to no modification.
+100% corresponds to 2X scaling of existing variations for maximum expressivity and exaggerated
pitch.
 DIALOGUE CONTOUR 112

Reset Curve
Removes all custom nodes from the curve, resetting it to default. Two nodes are present in the
default curve, one at the start and one at the end of the current selection. The default nodes are set
to 0 semitones (no pitch adjustment).

Curve Smoothing
Adjusts the amount of smoothing applied between nodes on the contour curve. Smoothing is a
global control and is applied to all nodes on the curve.
• Lower smoothing values: Applies little to no smoothing between nodes on the curve. Allows for
stricter transitions between nodes.

• Higher smoothing values: Applies more smoothing between nodes on the curve, resulting in
a gradual, rounded slope between points on the curve. Allows for more gradual transitions
between nodes.

Formant Scaling
Adjusts the amount of formant shift applied when adjusting pitch. The formants are shifted
together with pitch in the same or opposite direction. When the Formant Scaling is 0, the formants
stay in place. When it is 1, the formants move together with the pitch shift, i.e. the pitch shift does
not preserve formants.
This control can be helpful for maintaining or correcting the timbre and quality of the dialogue after
processing. In some situations, the formants may sound unnaturally high or low after processing.
Formant scaling can be used to correct for these unnatural sounding results.

Alternative Modules
For more generalized, non-dialogue specific pitch envelope editing, try using the Variable Pitch
module. The Variable Pitch module features the ability to shift pitch over the course of a selection
with or without preserving timing and is suitable for use on a wide range of input material.
 DIALOGUE ISOLATE 113

33. Dialogue Isolate

STD & ADV | Module & Plug-in

Overview
Dialogue Isolate is designed to separate spoken dialogue from a constant or non-stationary
background noise, such as hiss, crowds, traffic, footsteps, weather, or other noise with highly
variable characteristics. It can be particularly effective at increasing the level of dialogue in
challenging low signal-to-noise ratio recordings.
New in RX 11, Dialogue Isolate now includes reverb removal technology, which separates dialogue
from the characteristic echo and reverberation generated by different spaces, such as a tiled room,
a lecture hall, or a cathedral.
Dialogue Isolate uses a machine learning algorithm trained on a large library of speech, noise, and
reverberant data. It automatically detects and separates dialogue, noise, and reverberation into
three distinct signal components. The levels of the separated components can be independently
adjusted using the Voice, Reverb, and Noise gain controls.

Controls (STD and ADV)

Voice
Adjusts the amount of gain (in dB) applied to the components identified as voice.

Reverb
Adjusts the amount of gain (in dB) applied to the components identified as reverb.

Noise
Adjusts the amount of gain (in dB) applied to the components identified as noise.
 DIALOGUE ISOLATE 114

Sensitivity
Determines how much of the noise and reverb will be extracted from dialogue by the separation
algorithm.
• Higher values will instruct the separation algorithm to broadly define what it categorizes as
noise and reverb in the input signal. This can result in more significant noise reduction, at the
cost of introducing artifacts and potentially reducing dialogue clarity.
• Lower values will instruct the separation algorithm to narrowly define what it categorizes as
noise and reverb in the input signal. This can result in more noise being included in the rendered
signal, but can help to maintain dialogue clarity.

Quality

The following quality modes are available in the Dialogue Isolate module.
• Good / Real-time: A fast, high-quality processing mode that is optimized for speed (low-
latency) and for use in the plugin.
• Best / Offline: Our highest-quality processing mode with fewer artifacts and better isolation at
the cost of a longer processing time (higher latency). For use in the RX Audio Editor and for
offline bounces/renders in DAWs. (Advanced only)

The RX 10 Dialogue Isolate legacy processing modes do not exist in RX 11.

Controls (ADV only)

The following controls are only available in the Advanced version of the Dialogue Isolate module
and plugin (ADV):

An expand button shows or hides the metering and multi-band interface by making the window
larger or smaller.
 DIALOGUE ISOLATE 115

If the reduction of any band is set to a value other than 100%, the expand button will be highlighted
in blue to indicate that the multi-band controls will have an effect on the output.

Multi-band Interface
The multi-band interface in the Advanced module and the plugin adjusts the amount of processing
across four different frequency bands (Low, Low Mid, High Mid, High). Each band is individually set
from 0 to 100% of the processing parameter. (For example, if the Noise gain slider is set to -10dB,
and a band is set to 0%, the gain of the detected noise in that band will be 0dB. If the band is set
to 100%, the gain of the detected noise in that band will be -10dB.) The three crossover points are
adjustable.
Difference metering provides visual feedback as to how much gain adjustment is happening
across the four bands.

Alternative Modules
For stationary noise, such as hiss, buzz, line noise, etc., Dialogue Isolate should produce good
results, but we also suggest trying the Spectral De-noise module if you are having trouble
achieving acceptable results with Dialogue Isolate. Spectral De-noise is not focused on dialogue
and can be used for a variety of source signals, like singing voice, instrumental tracks, or full mixes.
For more general, non-dialogue specific reverb reduction, try using the De-reverb module.
 GUITAR DE-NOISE 116

34. Guitar De-noise

Module & Plug-in

Overview
Guitar De-noise is designed to provide control over acoustic and electric guitar performance-
related noises such as squeaks, picking sounds, and the hum or buzz that comes from guitar
pickups or amplifiers. With Guitar De-noise, you’ll be able to easily adjust these noises or eliminate
them completely, depending on the needs of the production.
Guitar De-noise consists of three sections, each targeting its own type of artifact. The Amp section
eliminates hum or buzz of amplifiers. The Squeak section attenuates fret noises. The Pick section
tames excessively sharp attacks of plucked strings.

Controls
The Guitar De-noise module is divided into three main sections: Amp , Squeak , and Pick.

Amp
Amp section is designed to deal with the hum or buzz that comes from guitar pickups or
amplifiers. Unlike the De-hum module, it can handle noises consisting of hundreds of harmonics
that reach well into the high-frequency range. It can also deal with buzzes that do not consist of a
single harmonic series: the Learn feature captures a snapshot of tonal noises at the arbitrary set of
frequencies.
The noise for the Amp section needs to be tonal and static. If noise contains both tonal and
broadband components (like hum + hiss), the broadband components can be addressed with a
separate pass of Spectral De-noise or Voice De-noise. Static noise means that the frequencies of
hum or buzz do not change in time.
 GUITAR DE-NOISE 117

• LEARN: will intelligently identify the noise profile of the hum or buzz in your audio
• In the plug-in:
• Loop a section of audio that contains only the hum or buzz that you want to reduce.
• Click Learn to start learning.
• After a moment click Learn again to finish the process.
• In the RX Application:
• Select the section of audio that contains only the hum or buzz that you want to reduce.
• Click Learn.
• Sensitivity: determines how much hum or buzz will be removed from the signal. Use higher
sensitivity values when the amplitude of noise varies in time: it will allow the algorithm to
attenuate noises that are louder than the learned snapshot. If the noise level is constant, using
lower sensitivity will allow for better preservation of the tone of the guitar.
• Resolution: sets the maximum number of harmonics that will be removed from the amp noise.
The default of 128 will be enough for most cases, but greater resolution is available at the cost
of increased CPU usage.

Squeak
The Squeak section is for controlling string noises (fret/finger squeaks) that occur when sliding up
or down the guitar neck between notes or chords.
 GUITAR DE-NOISE 118

• Sensitivity: determines how many string squeaks are detected in the signal. Increasing
sensitivity to higher amounts can attenuate some high-pitched guitar notes.
• Reduction: adjusts the amount of attenuation that is applied to the detected squeaks in the
signal.
• Duration:
• Short is good for the majority of shorter squeaks up to 200 ms in length.
• Long can handle both short and long squeaks up to 1000 ms in length. This mode has high
latency.

Pick
The Pick section allows you to reduce the attack of picked or plucked guitar strings for situations
where they might stand out too much in music or production sound.

• Sensitivity: determines how picks or plucks are detected in the signal. Increasing sensitivity to
higher amounts will impact the tone.
 GUITAR DE-NOISE 119

• Reduction: adjusts the amount of attenuation that is applied to the detected picks in the signal.
• Attack: determines how quickly the reduction is applied by altering the compressor’s attack
time.

Processing Tips
• Before learning the noise profile in the Amp section, identify and select the longest section
(ideally a few seconds in length) of the recording that contains only the hum or buzz you wish
to remove or reduce.
• Use the Ear button available on each section to set the sensitivity so that only the desired
noises are removed. This will help to preserve the original tone.
• Use the reduction amount sparingly at first and increase it to higher settings to remove more of
the noise if needed.

Alternative Modules
While the Amp section of Guitar De-noise successfully attenuates tonal noises, like hum or buzz,
it may not work well on broadband noises like hiss or rumble. We suggest trying RX’s Spectral
De-noise or Voice De-noise modules for efficient attenuation of broadband noises. Voice De-noise
has been specifically designed to provide high efficiency, zero-latency adaptive noise removal
when inserted on a track in your DAW or NLE. The Spectral De-noise plug-in is far more resource
intensive and has higher latency.
Although the Squeak section of Guitar De-noise is able to automatically identify and remove most
of the fret noises, sometimes a manual correction of a few outstanding squeaks is required. In
such cases, running Squeak with higher Sensitivity on a few isolated selections may be helpful. An
alternative way employs RX’s Spectral Repair module for precise editing of the squeaks in manual
selections.
When the flexibility of the Pick section is not enough, RX’s modules like De-ess (especially, the
Spectral algorithm) or De-click provide alternative automatic ways of attenuating sharp attacks.
Of course, in the most difficult situations, manual tools like Spectral Repair or Gain provide the
ultimate editing precision.
 INTERPOLATE 120

35. Interpolate
Overview
The Interpolate module is used for repairing individual clicks below 4000 samples in length.
This module corrects clicks by synthesizing a replacement signal based on the content in your
selection. It can be used as a substitute for the ‘pencil’ editing tool found in many audio editing
applications.

Controls

Quality
Adjusts the complexity of the sythesized replacement signal by adjusting the interpolation order.
This allows you to tailor the results of the processing to better fit the surrounding audio.
The following images illustrate the effect of Interpolate processing on a selection that is 214
samples in length. The images are zoomed in to illustrate the effect of adjusting the Quality
parameter.
INTERPOLATE 121
 LOUDNESS OPTIMIZE 122

36. Loudness Optimize

Overview
The Loudness Optimize module gives you information and control over your audio's loudness
measurement and the relative gate in the BS.1770 loudness algorithm. When momentary loudness
falls below this gate, it does not contribute to the file's overall integrated loudness measurement.
Applying upward compression can bring quiet content above the gate, causing it to contribute
to the measurement and reduce your audio's integrated loudness. As a result, your audio can
playback at a higher level on loudness-normalized streaming platforms.
There may also be cases where your audio's momentary loudness is already above the gate, and in
this case we hope that Loudness Optimize will be a valuable check to ensure that your audio will
sound optimally loud on streaming platforms.

Metering and Readouts

• LOUDNESS TRACE: Represents the momentary loudness of your audio over time. When audio
is above the LUFS Gate and contributes to the integrated loudness measurement, it is colored
blue. Audio below the LUFS Gate that does not contribute to the measurement is colored gray.
This chart can be scrolled and zoomed by dragging and using the mouse wheel on either axis.
The chart will focus on whatever region in time is highlighted in the main RX window. You can
also hover over the chart to see the audio's momentary loudness at any point in time.
• LUFS GATE LINE: The dashed white line represents the relative gate in the BS.1770 Integrated
Loudness measurement. This line will re-position dynamically depending on the audio and
cannot be directly adjusted by the user. Its position is 10 LU below the average momentary
loudness measurement.
• MEASURED %: Displays the amount of audio in the file that is above the LUFS Gate. Low
Measured % values indicate that there is potential for loudness optimization because there is
significant content that can be raised above the gate and therefore reduce integrated loudness.
As the Measured % approaches 100%, it is less likely that the audio's integrated loudness
measurement can be reduced.
 LOUDNESS OPTIMIZE 123

• INTEGRATED LUFS: Displays the integrated loudness of the audio file after applying the LUFS
gate. This is the measurement that streaming platforms will use when calculating loudness
normalization.

Controls

Boost
Controls the maximum amount of gain boost in dB that the processing will apply to quiet content.

Threshold
Determines the level in dBFS where the upward compression will begin to be applied. Audio above
this threshold level will be completely unaffected, so the audio's original peak values and limiting
will be maintained. Audio below this threshold will be upward-compressed with the specified Ratio,
subject to the maximum Boost. Threshold is also displayed in the main RX window (when the
waveform is visible) and can be adjusted via the connected handles.

Ratio
Affects how much compression is applied below the threshold. It works similarly to a typical
downward compressor's ratio control, but rather than affecting how much gain is reduced above a
threshold it affects how much gain is boosted below a threshold.
For example, a Ratio of 4:1 on a typical downward compressor means that for every 4 dB that
the signal exceeds the Threshold, the output gain will only be 1 dB above the Threshold. Applying
Loudness Optimize with a 4:1 Ratio means that for every 4 dB the signal falls below the Threshold,
the output will only be 1 dB below the Threshold.

Speed
Sets the ballistics of the upward compression, in milliseconds. It primarily determines how quickly
the processor will begin to boost when audio falls below the threshold. It also sets the lookahead
time for the upward compression. Longer speeds will set more lookahead time and be less prone
to distortion.

Learn
The Learn function will quickly apply many combinations of Boost and Threshold, measure the
resulting integrated loudness, and select a Boost and Threshold amount that will result in the most
optimal reduction in Integrated Loudness.
Learn will not change Ratio and Speed. Instead, your selected Ratio and Speed controls will be
accounted for as Learn is running. The following workflow is recommended:
1. Set your desired Ratio and Speed. The default Ratio and Speed of 4:1 and 120 ms are a good
starting point (achieve higher LUFS reduction with higher Ratios and faster Speeds).
2. Click Learn.
Learn will attempt to reduce the loudness of the highlighted audio region. In almost all cases, it is
recommended to run Learn on the entire audio file. Even if you only intend to apply the Loudness
Optimize processing to a specific section, the recommended workflow is to Learn from the entire
audio file and then highlight and render the learned settings to the specific section.
 LOUDNESS OPTIMIZE 124

It is possible that no reduction in integrated loudness can be achieved with any combinations of
Boost and Threshold, given the selected Ratio and Speed. In this case, the Loudness Optimize
module will show this message after running Learn:
 MOUTH DE-CLICK 125

37. Mouth De-click

Module & Plug-in

Overview
Mouth De-click detects and reduces mouth noises such as clicks and lip smacks. It’s designed for
use on longer audio selections, but it can also be used to remove individual clicks.

Controls

• SENSITIVITY: Determines how many mouth clicks are detected in the signal. Increasing
sensitivity can impact plosives, reducing or damaging the original signal.
• FREQUENCY SKEW: Targets the detection and removal of clicks to lower or higher frequencies.
Negative values are more suitable for generic clicks such as those found on vinyl recordings. A
setting of zero or above targets mouth clicks in the middle frequencies.
• CLICK WIDENING: Click Widening extends the repair area around detected clicks,
compensating for mouth sounds such as lip smacks that have a decay.

More Information
Running Mouth De-click twice sometimes produces a better result than a single pass. This is
because an initial run can miss the quieter clicks that were masked by louder clicks.
 MUSIC REBALANCE 126

38. Music Rebalance

Module & Plug-in

Overview
Music Rebalance leverages a machine learning algorithm trained to identify and separate the
following elements in a mix: Vocal, Bass, and Drums. Any content that is not otherwise identified
as Vocal, Bass, or Drums will be categorized as Other. This typically includes all the melodic
instruments. The level of each mix element can be independently adjusted after separation.
Music Rebalance can be useful for adjusting the level of a particular mix element when the original
tracks or stems are not available as an alternative solution. In some cases, it can also be used
to isolate a single mix element (e.g. the lead vocal) by reducing the level of the other three mix
elements. It is available as a module in the RX 11 Audio Editor, a VST-3 and AU ARA plug-in, and as
an Audiosuite and AAX real time plug-in in Pro Tools.

Controls

• VOCAL: Adjusts the level (in dB) of vocal signals.

• BASS: Adjusts the level (in dB) of bass signals.
• DRUMS: Adjusts the level (in dB) of drums and percussion signals.
• OTHER: Adjusts the level (in dB) of melodic instruments.
• MUTE BUTTON: Beside each of the separation sliders is a mute button for removing the
separated stem from what is being auditioned.
• SOLO BUTTON: Beside each of the separation sliders is a solo button for auditioning only that
stem.
• QUALITY: There are three separation algorithms to choose from in Music Rebalance: Good,
Better, and Best. Good provides separation results the fastest. Better improves separation
quality with only a minor decrease in processing speed. Best outputs the highest separation
quality, including anti-phasing processing, with the slowest speed.
 MUSIC REBALANCE 127

• SENSITIVITY: Individually adjusts the attention of the stem splitting algorithm toward a
particular stem. When Sensitivity is higher, more of the subtle signals will get classified as
this stem. For example, increasing the Sensitivity of vocals will preserve more of the breath
sounds in the vocal stem. However higher sensitivity may also cause some leakage of other
instruments into this stem. Note: raising the Sensitivity of all stems to 100% simultaneously
does not change the results of stem separation (as the algorithm cannot be more attentive to
everything) - Sensitivity only operates on relative values.

The RX 11 Music Rebalance Audiosuite plug-in does not include the ability to preview
processing before rendering.

Stem Split
Stem Split automatically splits your selected audio into four distinct stems (Vocal, Bass,
Percussion, Other), delivering each stem to its own tab. This process ignores the Gain settings
for individual stems and always assumes 0 dB of gain. Once the stems are separated, you can
process and export each stem as an independent audio file.

Getting started with Music Rebalance in Logic

Music Rebalance can be used in Logic as an ARA 2.0 plug-in (Rosetta only). In order to initialize the
ARA version of Music Rebalance, please follow the following steps:
1. Update to the latest version of Logic Pro for full compatibility with the ARA plug-in format.
There are known issues in versions prior to Logic Pro 10.7
2. Add the RX 11 Music Rebalance plug-in to a track.

3. Close and reopen Logic.

4. Add RX 11 Music Rebalance (ARA) as the first insert.

Still not seeing RX 11 Music Rebalance (ARA) appear after restarting? Make sure you
are adding it as the first insert on your track. Then try closing Logic and deleting
your AudioUnits cache here: ~/Library/Caches/AudioUnitCache, and repeat the steps
above.
 MUSIC REBALANCE 128

Using Music Rebalance (ARA) in Logic

Now that you have Music Rebalance (ARA) on your track, you’re ready to go.
1. Start playback to have Music Rebalance analyze your audio.
2. Once playback is started, you can pause playback and analysis will continue.

ARA plug-ins are updated when playback is started. If you make a change to your
audio, it will not be reflected until you’ve stopped and started playback again.

3. Once you have analyzed you’re audio, you’ll have access to the controls:

4. Move the sliders to rebalance your music in real time.The ARA plug-in uses the Good quality
mode. If you need even higher quality, use Music Rebalance in the RX 11 standalone editor.

Using Multiple Instances

When using multiple instances of the plug-in, the Music Rebalance window will follow the selected
track. The track name is shown at the bottom of the window.
 MUSIC REBALANCE 129

If you select a track that doesn’t contain Music Rebalance ARA, the plug-in will remember the last
selected track that contains an instance of Music Rebalance ARA.
If you have multiple Music Rebalance ARA plug-in windows open, they will all show the controls
and settings for the last selected track with Music Rebalance ARA.

Music Rebalance (ARA) Requirements & Limitations

• Music Rebalance (ARA) is supported in Rosetta mode only, as a limitation of Logic Pro
• Music Rebalance (ARA) must be the first plug-in on a track.
• Music Rebalance (ARA) cannot be added to a software instrument track or bus.
• Music Rebalance (ARA) is not supported by Selection Based Processing.
• Host presets are not supported by Music Rebalance (ARA).
• Automation is not supported by Music Rebalance (ARA).
• Playback must be started to gather audio data.
• After editing, playback must be stopped and started again to update data.
• Music Rebalance (ARA) cannot process Apple Loops, compressed audio, flex audio, or reversed
regions.
• These regions can be bounced in place in order to be processed by Music Rebalance (ARA).
 SPECTRAL DE-NOISE 130

39. Spectral De-noise

STD & ADV | Module & Plug-in

Overview
Spectral De-noise is designed to remove stationary or slowly changing tonal noise and broadband
hiss by learning a profile of the offending noise and then subtracting it from the signal. It can
be useful for tape hiss, HVAC systems, outdoor environments, line noise, ground loops, camera
motors, fans, wind, and complex buzz with many harmonics.
Spectral De-noise learns a profile of the background noise, then subtracts that noise when a
signal’s amplitude drops below the specified threshold. It is a flexible tool that can be used to
quickly achieve accurate, high-quality noise reduction. It also provides separate controls for tonal
and broadband noise, management of denoising artifacts, and an editing interface for controlling
reduction across the frequency spectrum.

Controls

Learn
When Learn is enabled, Spectral De-noise will capture a noise profile from your selection. After a
noise profile is captured using Learn, it remains fixed for the duration of processing. Manually
learned noise profiles are best suited to removing or reducing noise that is constant and
continuous throughout the duration of the file.
To learn a noise profile in Spectral De-noise:
1. Make a selection of the longest section of noise you can find in your file (ideally a few seconds
in length).
 SPECTRAL DE-NOISE 131

2. Click the Learn button to capture a noise profile:

• To capture a noise profile in the RX Audio Editor Spectral De-noise module, make a selection
and click “Learn”
• To capture a noise profile in the RX Spectral De-noise plug-in, engage the Learn button and
playback audio, OR choose “Preview” in Audiosuite to capture the noise profile from your
current selection.

MORE INFORMATION ABOUT LEARNING NOISE PROFILES

See the More Information section below to learn more about getting the best results
when capturing a noise profile and Learning a noise profile from multiple selections in
the RX 11 Audio Editor.

Adaptive Mode
When Adaptive Mode is enabled, the noise profile used for Spectral De-noise processing will
change based on the incoming audio. Adaptive mode can work well with noise sources that are
constantly changing, like recordings in outdoor environments, traffic noise, or ocean waves.

SPECTRAL DE-NOISE ADAPTIVE MODE PERFORMANCE NOTE

Adaptive mode in Spectral De-noise uses a significant amount of memory and
computational power. For a more efficient form of adaptive noise reduction, try the
Adaptive mode in Voice De-noise, which is designed to be highly efficient and zero-
latency.

Learning Time [s]

Determines the amount of lookahead time used by Adaptive mode when learning noise profiles
that change over time.

Threshold (Noisy/Tonal)
Controls the amplitude separation of noise and useful signal levels.
• Higher threshold settings reduce more noise, but also suppress low-level signal components.
• Lower threshold preserves low-level signal details, but can result in noise being modulated by
the signal. Threshold elevation can be done separately for tonal and random noise parts. A
good default is 0 dB.

TIP
If background noise changes in amplitude over time (like traffic noise or record
surface noise), raise the Threshold to accommodate for the changes.
 SPECTRAL DE-NOISE 132

Reduction (Noisy/Tonal)
Controls the desired amount of noise suppression in decibels. Spectral De-noise can automatically
separate noise into tonal parts (such as hum, buzz or interference) and random parts (such
as hiss). You can specify the amount of suppression for these parts separately (e.g. in some
situations it can be desirable to reduce only unpleasant buzz while leaving unobjectionable
constant hiss).

NOTE
Strong suppression of noise can also degrade low-level signals, so it is recommended
to apply only as much suppression as needed for reducing the noise to levels where it
becomes less objectionable.

Quality
Affects the quality and computational complexity of the noise reduction. This selection directly
affects CPU usage. RX’s Spectral De-noise module offers four algorithms that vary in processing
time.
• A: is the least CPU intensive process and has the lowest latency, which makes it the most
suitable for real-time operation. It reduces musical noise artifacts by time smoothing of the
signal spectrum.
• B: achieves more advanced musical noise suppression by using adaptive 2D smoothing (both
time and frequency). It is more CPU intensive and has more latency, but can still run in real-time
on some machines.
• C: adds multiresolution operation for better handling of signal transients and even fewer
musical noise artifacts. It is a very CPU intensive algorithm and is not recommended for
real-time operation.
• D: adds high-frequency synthesis for reconstruction of signal details buried in noise. The speed
of algorithm D is similar to algorithm C. This algorithm is not recommended for real-time
operation.

Artifact Control
Determines how much noise reduction will depend upon either spectral subtraction or wide band
gating.
• When using lower values, noise reduction will rely on spectral subtraction. This can more
accurately separate noise from the desired audio signal, but can produce musical noise
artifacts, resulting in a “chirpy” or “watery” sound during heavy processing.
• When using higher values, the noise reduction will rely more heavily upon wider band gating
which will have fewer musical noise artifacts, but sound more like broadband gating, resulting
in bursts of noise right after the signal falls below the threshold.

Noise Spectrum Display

The Noise Spectrum display shows useful information during both playback and when the noise
reduction process is being applied.
 SPECTRAL DE-NOISE 133

• Noise Spectrum Color Legend

• Input (Gray): spectrum of input audio signal
• Output (White): spectrum of the denoised output audio signal
• Noise Profile (Orange): the learned noise profile plus offset from the Threshold control
• Residual Noise (Yellow): desired noise floor after denoising, can be controlled by modifying
the Reduction Curve
• Reduction Curve (Blue): manual weighting of the noise reduction across the spectrum

Smoothing
When the Reduction Curve is enabled, this controls the amount of interpolation between your
reduction curve points, allowing for sharper or more gradual slopes between edit curve points.

Reduction Curve
When enabled, allows for fine tuning of the reduction spectrum with up to 25 edit points. This
enables you to customize the amount of noise reduction being applied across different frequency
regions.
• Higher edit point values result in less noise reduction in the associated frequency region.
• Lower edit point values result in more noise reduction in the associated frequency region.
• For example, if you wanted to reduce some low HVAC rumble but preserve some energy in
higher frequencies, you could drag the curve’s leftmost point down a little bit, then create a
point around 5 kHz and drag it up a bit.
You can interact with the Reduction Curve Edit points in the following ways:
▶ Left-click to add an end point, displayed as gray box along envelope curve.
▶ Right-click or drag it outside the screen to emove an edit point.
▶ Hold Shift while dragging reduction curve points to lock them on an axis.
▶ Hold Control/Command to get very fine control over positioning.

Reset
Returns the Noise Reduction Curve to its default setting of 0dB.

Advanced Settings
 SPECTRAL DE-NOISE 134

Algorithm Behavior (Advanced Settings)

Smoothing - Advanced
Controls the reduction of musical noise artifacts which can be a result of heavy denoising.
What is musical noise?
Musical noise is caused by random statistical variations of noise spectrum that cause random
triggering of sub-band gates. These artifacts are sometimes described as “chirpy” or “watery”
sounds left behind during the noise reduction process.

Algorithm
Selects the smoothing algorithm for the removal of random ripples (“musical noise” artifacts) that
can occur in the spectrogram when processing your audio. The strength of smoothing is controlled
by the Smoothing slider.
• SIMPLE: Performs independent noise gating in every frequency channel of FFT. Release time of
sub-band gates is controlled by the Release slider. This is a fast algorithm with low latency that
is suitable for real-time operation.
• ADVANCED & EXTREME: Perform joint time-frequency analysis of the audio signal which
results in better quality and fewer “musical noise” artifacts. These algorithms have higher
latency and computational complexity.

FFT Size (ms)

Selects the time and frequency resolution of the processing.
• Higher FFT sizes give you more frequency bands allowing you to cut noise between closely
spaced signal harmonics, or cut steady-state noise harmonics without affecting adjacent
signals.
• Lower FFT sizes allow for faster response to changes in the signal and produce fewer noisy
echoes around transient events.

RE-LEARN YOUR NOISE PROFILE IF YOU CHANGE FFT SIZE

If the FFT size is changed, it is recommended that you run the De-noise module’s
Learn feature again because the old noise profile was taken at a different FFT size and
therefore becomes inaccurate.
 SPECTRAL DE-NOISE 135

Multi-Res
Enables multi-resolution processing for the selected algorithm type. When you select the Multi-res
checkbox, the signal is analyzed in real-time and the most appropriate FFT size is chosen for each
segment of the signal. This is done to minimize the smearing of transients and at the same time
achieve high frequency resolution where it is needed.

NOTE
The FFT size control does not have any effect in multi-resolution mode as the FFT
resolution is selected automatically. The noise profile does not need to be re-learned
when switching to multi-resolution mode.

Fast Fourier Transform (FFT)

A procedure for the calculation of a signal frequency spectrum. The greater the FFT size, the
greater the frequency resolution, i.e., notes and tonal events will be clearer at larger sizes. However,
when using FFT-based processing, the more audio you remove from your source, the more likely
you are to create undesirable artifacts.

Noise Floor (Advanced Settings)

• Synthesis: Synthesizes high frequency material after denoising.

• When Synthesis is set to a value greater than zero, signal harmonics are synthesized after
denoising. The synthesized harmonics remain at the level of the noise floor, and serve to fill
in gaps in high frequencies caused by processing.
• Increasing Synthesis can increase the sense of life and air in processed audio. Too much
Synthesis may cause apparent distortion in the signal.
• ENHANCEMENT: Enhances signal harmonics that fall below the noise floor.
• Enhancement predicts a signal’s harmonic structure and places less noise reduction in
areas where possible signal harmonics could be buried in noise. This aids in preserving
high-frequency signal harmonics that may be buried and not detected otherwise.
• Enhancement can make the resulting signal brighter and more natural sounding, but high
values of harmonic enhancement can also result in high-frequency noise being modulated
by the signal.
 SPECTRAL DE-NOISE 136

• MASKING: Reduces the depth of noise reduction where you wouldn’t perceive any effect from
it.
• Masking enables a psychoacoustic model that dynamically controls suppression amount to
facilitate the use of softer suppression where noise is subjectively inaudible. When noise
in certain regions is calculated to be inaudible, this feature prevents any signal processing
in these regions. This potentially reduces the amount of processing done to the signal and
may positively affect overall signal integrity. The position of the slider controls the influence
of psychoacoustic model on suppression levels.
• If you need to cut very high, inaudible frequencies, set this to 0. Otherwise, leave this at 10.

NOTE
When the Masking slider is set to 0, the feature is turned off, and the amount
of noise suppression is uniformly governed to the yellow curve in spectrum
analyzer (more precisely — by the difference between the yellow curve and
orange curve).

• WHITENING: Shapes the noise floor after processing to be more like white noise. Whitening
modifies the amount of noise reduction (shown by the yellow curve) applied at different
frequencies to shape the spectrum of the residual noise.
• When Whitening is set to zero, the suppression is uniform at all frequencies, as controlled by
Reduction (tonal/broadband) sliders, and the suppressed noise has a similar spectral shape
to the original noise.
• When Whitening is set to the maximum value, the desired shape of suppressed noise floor
is made close to white noise, so that residual noise has more neutral sound.

UNDERSTANDING THE EFFECT OF THE WHITENING CONTROL

Changing the noise floor balance with Whitening can help prevent gaps from
over-processing, but an unnaturally white noise floor can introduce problems
like noise modulation when editing or mixing with other noises from a unique
space (like a set location.)

Dynamics (Advanced Settings)

• KNEE: Controls how surgical the algorithm’s differentiation is between the signal and noise.
This slider controls the sharpness of the gating knee in the denoising process.
• At higher values, transitions in the De-noise are more abrupt and can become prone to
errors in the detection of the signal with respect to the noise.
• At lower values, the denoising becomes more forgiving around the knee, and applies less
attenuation to signals that are only slightly below the threshold. This may result in a lower
depth of noise reduction, but can also have fewer artifacts.
 SPECTRAL DE-NOISE 137

• RELEASE [ ms ]: Selects the release time of sub-band noise gates in milliseconds. Longer
release times can result in less musical noise, but may also reduce or soften the signal’s initial
transients or reverb tails after the signal’s decay.

NOTE
The Release control is only available when the Simple algorithm is selected.

More Information

Tips for getting the best results when learning noise profiles manually
• Before learning a noise profile, identify and select the longest section (ideally a few seconds in
length) of the recording that contains only the noise you wish to remove or reduce.
• To ensure the best results, your selection should not contain any content that you wish to
preserve (for example, do not include any audio you do not consider to be “noise” in your
selection.)
• Usually you can find noise only sections at the beginning or end of a file, or during a pause or
break in the recording (for example, a pause between words in dialogue recordings.)

Learning a Noise Profile From Multiple Selections

In the RX standalone application, it is possible to create a spectral profile from multiple isolated
selections. This is useful when you have a file where it’s impossible to find enough isolated noise to
build the profile.
For example, if you are trying to restore a file where someone is speaking over noise, you can
select noise in frequencies where none of the voice is present at a given time. If you select enough
of this noise with the Lasso or Brush selection tools, you can create an accurate noise profile that
will give you good results with Spectral De-noise. You can create more than one selection at a time
by holding Shift while making a selection.
Select noise anywhere you can to build a better noise profile.
This feature is not available in the Spectral De-noise plug-in because it requires using RX’s spectral
selection tools as well as accurate calculation of the time and frequency of the selected areas.
If you are unable to create a full noise profile with multiple selections, RX can try to build a
reasonable noise profile out of your existing profile. If you have an incomplete noise profile, RX will
ask you if you want it to complete the profile.
For example, if you can only capture a low frequency rumble below 100 Hz, some broadband noise
between 200 Hz and 5000 Hz, and all the noise above 8000 Hz, RX can fill in the gaps for you.
Building a profile from multiple selections gives you some flexibility, and RX will guess any noise
you missed.
 SPECTRAL REPAIR 138

40. Spectral Repair

STD & ADV

Overview
Spectral Repair intelligently removes undesired sounds from a file with natural-sounding results.
This tool treats selections within the spectrogram/waveform display as corrupted audio that will be
repaired using information from outside of the selection. Select the noise you want to repair and
Spectral Repair will reduce it to the level of the noise floor, replace it with audio from around the
selection, or generate entirely new audio to fit the selection.

Attenuate
This mode removes sounds by comparing the content inside of a selection to the content
outside of the selection. Attenuate reduces spectrogram magnitudes in the selected area to match
magnitudes from the surrounding area, resulting in the removal of the sound without leaving an
audible gap behind. Attenuate does not resynthesize any audio. It modifies dissimilar audio in your
selection to be more similar to the surrounding audio.
Attenuate is suitable for recordings with background noise or where noise is the essential part
of music (drums, percussion) and should be accurately preserved. It’s also good when unwanted
events are not obscuring the desired signal completely. For example, Attenuate can be used to
bring noises like door slams or chair squeaks down to a level where they are inaudible and blend
into background noise.

Controls

• BANDS: Selects the number of frequency bands used for interpolation.

• A higher number of bands can provide better frequency resolution, but also requires wider
surrounding area to be analyzed for interpolation.
• A lower number of bands is ideal for processing short selections or transient signals.
• SURROUNDING REGION LENGTH: Defines how much of the surrounding content will be used
for interpolation.
 SPECTRAL REPAIR 139

• STRENGTH: Adjusts strength of attenuation.

• MULTI-RESOLUTION: The multi-resolution mode allows for better frequency resolution for
the interpolation of low-frequency content and better time resolution for the interpolation of
high-frequency content.
• BEFORE/AFTER WEIGHTING: Gives more weight to the surrounding audio before or after the
selection.
• DIRECTION OF INTERPOLATION: Determines where the material used in the repair process is
located in relation to the current selection.
• HORIZONTAL: Signal to the left and right of the current selection will be used for
interpolation.
• VERTICAL: Signal above and below the current selection will be used for interpolation.
• 2D: Signal above, below, to the left and to the right of the current selection will be used for
interpolation.

NOTE: DIRECTION OF INTERPOLATION CONTROL AVAILABILITY

Replace, Pattern and Partials + Noise tabs only utilize Horizontal mode and do
not display this option.

Replace
The Replace tab can be used to replace badly damaged sections (such as gaps) in tonal audio. It
completely replaces the selected content with audio interpolated from the surrounding data.

Replace Controls

• BANDS: Selects the number of frequency bands used for interpolation.

• A higher number of bands can provide better frequency resolution, but also requires wider
surrounding area to be analyzed for interpolation.
• A lower number of bands is ideal for processing short selections or transient signals.
• SURROUNDING REGION LENGTH: Defines how much of the surrounding content will be used
for interpolation.
• MULTI-RESOLUTION: The multi-resolution mode allows for better frequency resolution for
the interpolation of low-frequency content and better time resolution for the interpolation of
high-frequency content.
 SPECTRAL REPAIR 140

• BEFORE/AFTER WEIGHTING: Gives more weight to the surrounding audio before or after the
selection.

Pattern
This mode finds the most similar portion of the surrounding audio and uses this to replace the
corrupted audio. Pattern mode is suitable for badly damaged audio with background noise or for
audio with repeating parts.

Pattern Controls

• BANDS: Selects the number of frequency bands used for interpolation.

• A higher number of bands can provide better frequency resolution, but also requires wider
surrounding area to be analyzed for interpolation.
• A lower number of bands is ideal for processing short selections or transient signals.
• SURROUNDING REGION LENGTH: Defines how much of the surrounding content will be used
for interpolation.
• MULTI-RESOLUTION: The multi-resolution mode allows for better frequency resolution for
the interpolation of low-frequency content and better time resolution for the interpolation of
high-frequency content.
• PATTERN SEARCH RANGE: Selects the length of the audio segment used in a search for a
suitable replacement interval. For example, setting it to 5 seconds will allow search within ±5
second range from the selection.

Partials and Noise

The advanced version of Replace mode. It restores harmonics of the audio more accurately with
control over the Harmonic sensitivity parameter.
This mode allows for higher-quality interpolation by explicit location of signal harmonics from the 2
sides of the corrupted interval and linking them together by synthesis.
Partials + Noise is able to correctly interpolate cases of pitch modulation, including vibrato. The
remaining of non-harmonic material (“residual”) is interpolated using Replace method.
 SPECTRAL REPAIR 141

Partials and Noise Controls

• BANDS: Selects the number of frequency bands used for interpolation.

• A higher number of bands can provide better frequency resolution, but also requires wider
surrounding area to be analyzed for interpolation.
• A lower number of bands is ideal for processing short selections or transient signals.
• SURROUNDING REGION LENGTH: Defines how much of the surrounding content will be used
for interpolation.
• HARMONIC SENSITIVITY: Adjusts amount of detected and linked harmonics.
• Lower values will detect fewer harmonics
• Higher values will detect more harmonics and can introduce some unnatural pitch
modulations in the interpolated result.
• MULTI-RESOLUTION: The multi-resolution mode allows for better frequency resolution for
the interpolation of low-frequency content and better time resolution for the interpolation of
high-frequency content.
• BEFORE/AFTER WEIGHTING: Gives more weight to the surrounding audio before or after the
selection.

Surrounding Region Display

• SURROUNDING REGION SHADING: When using the Spectral Repair module, your selections
will be shown with a dotted line surrounding your selected region. This dotted line is directly
controlled by the Surrounding Region and Before/After Weighting controls inside of your
Spectral Repair modules, and provides a visual representation of your set values.
• The surrounding region is the region that RX uses for interpolation of the selected region.
The data from the surrounding region is used to restore the selected region.

Workflow

Applying Spectral Repair

1. To start working with Spectral Repair, switch to the spectrogram view by dragging the
spectrogram/waveform transparency balance slider to the right.
 SPECTRAL REPAIR 142

2. Next, identify the unwanted event on a spectrogram and select it using a selection tool (use the
time-frequency, brush, lasso or magic wand tools to fit your selection to the corrupted audio)
3. You can audition just the signal in your selection by pressing the Play Selection button in the RX
transport.
4. Once you’ve found the event(s) to repair, select the appropriate Spectral Repair mode from the
tabs at the top of the Spectral Repair settings window.
5. You can use the Compare Settings functionality to audition the processing before applying it, or
click the process to apply the active Spectral Repair tab’s settings.

More information
This section contains useful information, examples and tips for getting the most out of the
Spectral Repair module.

Visual example
The following image shows the selection before processing on the top and the selection after
processing with Spectral Repair on the bottom.

Processing limitations
Depending on the mode and settings, Spectral Repair will have varying limits to the amount of
audio that can be processed in your selection.
• Unlimited — Attenuate when in Vertical mode only.
• 10 seconds — Attenuate Horizontal or 2D; Replace modes.
• 4 seconds — Pattern, Partials + Noise modes.
• Longer selections will automatically adjust processing to use the correct mode.

Spectral Repair as an alternative to De-click processing

When used with a time selection, Spectral Repair is able to provide higher quality processing than
De-click for long corrupted segments of audio (above 10 ms).
 SPECTRAL REPAIR 143

Example use cases for Spectral Repair

When used with a time/frequency, lasso, brush, or wand selection, it can be used to remove
(or attenuate) unwanted sounds from recordings, such as: squeaky chairs, coughs, wheezes,
burps, whistles, dropped objects, mic stand bumps, clattering dishes, mobile phones ringing,
metronomes, click tracks, door slams, sniffles, laughter, background chitchat, digital artifacts from
bad hardware, dropouts from broken audio cables, rustle sounds from microphone movements,
fret and string noise from guitars, ringing tones from rooms or drum kits, squeaky wheels, dog
barks, jingling change, or just about anything else you could imagine. Spectral Repair can also
effectively repair gaps or replace audio intelligently by using advanced resynthesis techniques.

Increase efficiency with the Find Similar Event Tool

Some unwanted events consist of several separate regions on a spectrogram. In some cases,
it’s possible to achieve more accurate results by repairing several smaller selections one by one,
instead of one large selection. You can use the Find Similar Event tool to save time when searching
for and fixing many similar events in large files.

Use the Compare Settings window to experiment with Spectral Repair

processing
Sometimes it’s worth trying several different methods or number of bands to achieve the desired
result. A higher number of bands doesn’t necessarily mean higher quality! We encourage you to
use the Compare Settings window to experiment and find the best settings for the project at hand.

Adjust Surrounding Region Length and Before/after weighting to perfect

your processing
Common parameters for many modes include Surrounding region length which determines how
far around the selection Spectral Repair will look for a good signal. Before/after weighting allows
you to use more information from either before or after the selection for interpolation. For example,
if your unwanted event is just before a transient (such as a drum hit) in the audio, you may want
to set this parameter to use more of the audio before the selection to prevent smearing of the
transient into the selection.
 SPECTRAL RECOVERY 144

41. Spectral Recovery

ADV

Overview
Spectral Recovery automatically adds missing frequencies to bandwidth-limited speech content
and patches holes in the frequency spectrum due to compression artifacts. Spectral Recovery is
particularly useful in handling VoIP recordings, from Skype or Zoom, where there is a hard cutoff
above which no audio exists and the natural low end may be filtered or reduced.

Controls

Learn
When Learn is pressed, Spectral Recovery analyzes your selection to determine the suggested
values for the high and low cutoff frequencies.

Spectral Patching
When selected, this automatically fills in holes in the spectrogram between the high and low cutoff
frequencies by sampling the area around the missing audio. This control is very subtle audibly, but
you can see the effect in the Spectrogram.

Low Gain
This sets the level of synthesized low frequency signals below the cutoff.
 SPECTRAL RECOVERY 145

Low Cutoff
Spectral Recovery synthesizes audio below this frequency and fills the spectral holes above this
frequency.

TIP
Try adjusting the low cutoff to just above the fundamental frequency of your speaker
for Voice-over-IP material to make it sound more natural.

High Cutoff
Spectral Recovery synthesizes audio above this frequency and fills the spectral holes below this
frequency.

High Gain
This sets the level of synthesized high frequency signals above the cutoff.

Render
To apply Spectral Recovery processing to the current selection, click the Render button in the
bottom right hand corner of the window.
 STREAMING PREVIEW 146

42. Streaming Preview

Overview
The Streaming Preview module applies the same loudness normalization and lossy encoding that
streaming platforms will apply. This means you can use Streaming Preview to hear how your song
will sound on those platforms and compare it with reference songs. We hope this process helps
you uncover improvements for your mixes and masters and gives you more confidence to release
your music.

Controls

• PLAYBACK LOUDNESS: Represents the loudness normalization target for the streaming
platform in Loudness Units Full Scale (LUFS). When Streaming Preview is applied, gain will
be applied to your audio to match this loudness. The loudness measurement and normalization
occurs after lossy encoding has been applied.
• APPLY POSITIVE GAIN: When this option is enabled and your audio is quieter than the
Playback Loudness you're targeting, positive gain will be applied to bring your audio up to
the target. However, streaming platforms like Spotify will not boost your audio to be greater
than -1 dB True Peak, so the same is true in the Streaming Preview module. For example, if an
audio file's loudness is -20 LUFS, and its True Peak maximum is -5 dB FS, Streaming Preview
will only lift the track up to -16 LUFS.
• CODEC: Selects the lossy encoding format that streaming platforms apply to reduce the
amount of data required to stream and store audio.
• LOSSLESS (NONE): Does not apply any lossy compression to simulate streaming platforms
offering HiFi lossless audio like Tidal and Apple Music.
• OGG: A free, open container format maintained by the Xiph.Org Foundation. It is primarily
used by Spotify.
• MP3: (MPEG Audio Layer III) is a coding format for digital audio developed largely by the
Fraunhofer Society in Germany.
• AAC: (Advanced Audio Coding) is designed to be the successor of the MP3 format and
generally achieves higher sound quality than MP3 at the same bit rate. It is used when
streaming lossy audio through Apple Music and Tidal.
• OPUS: Opus is a lossy audio coding format developed by the Xiph.Org Foundation and
standardized by the Internet Engineering Task Force. It’s designed to efficiently code speech
and general audio in a single format, while remaining low-latency enough for real-time
interactive communication and low-complexity enough for low-end embedded processors.
It is the primary codec used by YouTube.
 STREAMING PREVIEW 147

• QUALITY: Selects the bitrate for the lossy encoding. This represents the amount of information,
or detail, that is stored per unit of time of a recording. Lower bit rates will result in more audio
loss and a reduction in sonic quality. When the Lossless codec option is selected, the Quality
control is not accessible.

Workflow and Presets

The recommended workflow with the Streaming Preview module is to use the provided presets.
We have researched the major streaming platforms and created these presets to represent their
Playback loudness levels and codecs. Some streaming platforms enable you to make changes that
will affect these settings and these options are also represented in the presets. Streaming platform
companies may change their settings and we will do our best to keep the presets up-to-date if the
settings change.
Streaming Preview is designed only for auditioning codec processing/loudness normalization.
We do not recommend exporting or uploading your audio to a streaming service after applying
Streaming Preview processing. This would effectively bake in any lossy encoding you hear in the
Preview.
 VOICE DE-NOISE 148

43. Voice De-noise

Module & Plug-in

Overview
Voice De-noise is an intuitive, zero latency de-noiser that offers high quality results on a variety
of material. Voice De-noise can intelligently analyze speech signals and determine the best noise
threshold for your signal. In a DAW, this feature can be used to write automation in case you need
to override the automatic settings and correct the noise threshold by hand.
How does Voice De-noise processing work?
• Under the hood is a series of 64 psychoacoustically spaced bandpass filters which act as a
multiband gate to pass or stop a signal based on user-defined threshold values.
• If a signal component is above the threshold for the filter, it will be passed (not processed).
• If a signal component is below the threshold for the filter, it will be attenuated (processed).

Controls

• ADAPTIVE MODE: Analyzes the incoming signal and adjust the noise threshold automatically
to compensate for changes in the noise floor. This can be useful for removing noise from
recordings with variable noise floor and continual noisy sections, and works well for almost any
recording of dialogue and spoken word.
• The noise threshold settings in Adaptive Mode may be different from the settings achieved
by running Learn to set the noise threshold manually.
• Because the adaptive noise threshold is continually being adjusted, it is set lower to prevent
artifacts from occurring.
• LEARN: When using Manual mode, you can use the Learn button to set the noise threshold to a
noise reference.
• Find a passage of pure noise in your audio and use Learn to analyze it.
• Longer selections of noise will set the Threshold Nodes to more ideal locations.
• We recommend finding at least one second of pure noise to Learn your noise profile from.
 VOICE DE-NOISE 149

• OPTIMIZE FOR DIALOGUE OR MUSIC: Because dialogue tends to be in short bursts and vocals
tend to have sustained notes, we’ve added modes to provide better results when applying Voice
De-noise processing.
• Optimize for DIALOGUE reacts to noise changes faster and isn’t meant to handle the noise
found in sung vocals.
• Optimize for MUSIC does not attenuate sustained notes and is more transparent when
applied to sung vocals.
• FILTER TYPE: Choosing the filter type changes the bandwidth of the noise reduction filters.
• SURGICAL Mode: Removes more unwanted noise than Gentle Mode, but makes some
sacrifices in terms of timbre and can lead to musical noise artifacts. Musical noise is
caused by random statistical variations of noise spectrum that cause random triggering of
sub-band gates. These artifacts are sometimes described as “chirpy” or “watery” sounds left
behind during the noise reduction process.
• GENTLE Mode: provides more transparent noise reduction than Surgical, but removes less
high end “sizzle.”
• THRESHOLD NODES: The Threshold Node controls on the frequency spectrum display allow
you to modify the noise threshold curve, which can be thought of as the “noise profile.” These
six points can be adjusted manually to suit the noise currently in your signal. These controls
can be automated to compensate for shifts in the audio’s noise floor.
• In ADAPTIVE Mode, the Threshold Nodes are adjusted automatically in real-time.
• In MANUAL mode, more than one Threshold Node can be selected at a time for manual
adjustment by clicking and dragging anywhere on the interface.
• THRESHOLD: The master Threshold control allows you to offset all Threshold Node values by
the same amount. If you find that processing is too aggressive or processing is affecting audio
you want to leave unprocessed, try adjusting this control.
• REDUCTION: Provides control over the maximal depth of noise reduction (in dB) that will occur
per frequency band while a signal component is below its threshold. If you have your thresholds
set properly and don’t like the results you’re getting, try adjusting this control.
• METERING
• The Input Spectrum meter shows the level of the signal at the input of the denoiser filters.
• The Output Spectrum meter shows the level of the signal at the output of the denoiser
filters.
• The Gain Reduction Region is the area between the Input and Output Spectra. This shows
the amount of noise reduction processing being applied to your signal.

Voice De-noise Plug-in

Voice De-noise has been specifically designed to provide high efficiency, zero latency adaptive
noise removal when inserted on a track in your DAW or NLE. The Spectral De-noise plug-in is far
more resource intensive and uses higher latency.
 WOW & FLUTTER 150

44. Wow & Flutter

ADV

Overview
The Wow & Flutter module allows for transparent correction of pitch variations in your audio due
to inconsistent speed of the recording medium. Wow & Flutter can be tailored to different rates of
wow, and allows for global pitch correction to compensate for static pitch offsets. The detected
wow or flutter pitch contour can also be written to the spectrogram for diagnostic purposes.

NOTE
When applied to a bandwidth-limited selection, Wow & Flutter will only use the
selected frequency band for detection of pitch variations. When applied to a full-band
selection, the algorithm will automatically determine the most suitable frequency
range for analysis. Multiple selections will be combined into one time selection and
processed as a whole.

Controls

Wow
Addresses slower pitch variations (typically 0 to 5 Hz).
 WOW & FLUTTER 151

You can choose from the following Wow Rate options:

• Fast: targets wow in the 2 to 8 Hz range.
• Medium: targets wow in the 0.5 to 2 Hz range.
• Slow: targets wow in the 0 to 0.5 Hz range.
When this mode is selected, a Display wow checkbox is available in the module footer. See the
Display Wow/Display Flutter section below for more information.

Flutter
Addresses quicker pitch variations (8 to 40 Hz).

When this mode is selected, a Display Flutter checkbox is available in the module footer. See the
Display Wow/Display Flutter section below for more information.

Sensitivity
Determines how aggressively the pitch variations will be corrected. A lower sensitivity can prevent
desirable pitch fluctuations like vibrato from being impacted.

Center Global Pitch

When checked, this control centers the audio around the specified concert pitch. If a static pitch
offset is present in the material, it will be corrected toward a specific tuning of the musical scale.
The default pitch value is 440 Hz.
When unchecked, processing will preserve the length of the selection and its average pitch intact.
 WOW & FLUTTER 152

Display Wow/Display Flutter

Enabling this checkbox will render the detected pitch contour of the wow or flutter onto the audio
file. No correction will be applied, but the detected pitch variations will be printed as a diagnostic
tool.

Rendered with Display Wow checked Rendered with Display Flutter checked

NOTE
Wow & Flutter is not available in the Batch Processor, Module Chain or Composite
View.
 AZIMUTH 153

45. Azimuth
ADV

Overview
The Azimuth module provides control over left and right channel gain and delay. Azimuth
adjustment can help repair stereo imbalances and phase issues that can occur as a result of
issues introduced by speed inconsistencies.

Controls

• LEVEL (dB): Adjusts the gain of the left and right audio channels
• ADAPTIVE MATCHING: Enables automatic gain adjustment of the right channel in order to
match the level of the left channel over time.
• DELAY (samples/ms): Allows for manual adjustment over the delay in samples or milliseconds
of the left and right audio channels. For very accurate azimuth correction, RX uses
oversampling to achieve sub-sample delays.
• ADAPTIVE AZIMUTH ALIGNMENT: Enables automatic time-variable adjustment of the right
channel’s sample delay in order to align the waveform with the left channel.
• SUGGEST: Analyzes the selection and determines the appropriate amounts of fixed gain and
delay to apply in order to align the two channels.

Suggest and Adaptive modes are only available on stereo files

The Suggest function and Adaptive Matching modes are meant to function on stereo files, these
controls will be disabled in the Azimuth module interface when a mono file is selected.
 AZIMUTH 154

More Information
• Use Azimuth to fix gain and delay alignment issues:
• Azimuth adjustment can be useful to repair inconsistent gain or delay alignment between
left and right channels.
• For example, gain and delay alignment inconsistencies can be introduced by improper tape
head alignment.
• Use Azimuth before processing with Center Extract: Azimuth adjustment is recommended to
be applied before Center Extract processing to achieve the best results.
 DITHER 155

46. Dither

Overview
Dithering is a necessary process when converting audio from a higher bit resolution to a lower bit
resolution. Dithering is used to tame the quantization distortion that happens when converting
between bit depths due to requantization. Dither also preserves more of the dynamic range
of a signal when converting to a lower bit depth. The Dither module applies iZotope’s MBIT+
dithering and noise shaping technology to maintain the highest audio quality possible when you
are converting to 24, 20, 16, 12, or 8 bits. MBIT+ uses psychoacoustic methods to distribute
dithering noise into less audible ranges. The result is a more pleasing sound and smoother fades.

Controls

New Bit Depth

This sets the target resolution (bit depth) of the audio file.

Noise Shaping
Sets the aggressiveness of dither noise shaping. It is possible to provide more effective and
transparent dithering by shaping the dithered noise spectrum so less noise is in the audible
range and more noise is in the inaudible range. You can control the aggressiveness of this
shaping, ranging from None (no shaping, plain dither) through Ultra (roughly 14 dB of audible
noise suppression). More noise shaping can cause slightly higher peaks in your signal, even at high
bit depths.

Dither Amount
The dithering amount can be varied from None (noise shaping only) to High.
• In general, the Normal dither amount is a good choice.
 DITHER 156

• No dithering or Low dither amount can leave some non-linear quantization distortion or dither
noise modulation, while higher settings completely eliminate the non-linear distortion at the
expense of a slightly increased noise floor.
• A dither amount setting of High with no noise shaping produces a standard TPDF dither: a
common white noise generation method with a triangular distribution of amplitudes between
−1 and +1 of the Least Significant Bit (LSB).

Auto-blanking
Automatically mutes dither output (i.e. dither noise) based on the selected mode and
characteristics of the input signal. Auto-blanking mode options include:
• IF QUANTIZED: Mutes dither output when already dithered or quantized signal is detected.
When existing quantization is detected, the auto-blanking feature will mute dither until a signal
with no existing dither is detected.
• ON SILENCE: Mutes dither output when silence is detected.
• OFF: Disables auto-blanking.

Limit Noise Peaks

Dither noise is random in nature and has a very low amplitude. However, after noise shaping,
especially in aggressive dithering modes like Ultra, the high-frequency dither noise is significantly
amplified, and the overall dither signal can show spurious peaks up to −60 dBFS during 16-bit
quantization. If such high peaks are undesirable, you can enable this option to effectively suppress
the spurious peaks in the noise-shaped dither.

Suppress harmonics
If, for some reason, any dithering noise is undesirable, simple truncation remains the only choice.
Truncation results in harmonic quantization distortion that adds overtones to the signal and
distorts the timbre. In this case you can enable Suppress Harmonics option to slightly alter the
truncation rules, moving the harmonic quantization distortion away from overtones of audible
frequencies. This option doesn’t create any random dithering noise floor. Instead it works more like
truncation, but with better tonal quality in the resulting signal. This option is applicable only in the
modes without dithering noise and without aggressive noise shaping.
 EQ 157

47. EQ

Overview
RX includes an eight-band parametric EQ module with six adjustable notch/shelving filters and two
adjustable passband filters. The EQ module is useful for manually shaping the overall sound of a
file or selection, for both corrective and enhancement purposes. Beyond a traditional equalizer, the
EQ also offers remarkably high Q values for precision filtering.
EQ can often be a simple first step to preparing a file for restoration, and can be used for cutting
harsh high frequencies, removing rumble from dialogue, steeply high passing out wind noise from
a location recording, or cutting distortion overtones to increase the intelligibility of a voice.

Controls

EQ TYPE
• Analog (IIR): a minimum-phase EQ with analog-like EQ curves.
• Digital Linear Phase (FIR): a linear-phase EQ with surgical EQ curves.

FREQUENCY PRECISION
This controls the number of bands, i.e. frequency resolution vs. time resolution, for the FIR filter.
A lower value will increase the frequency resolution, for tighter cuts, at the expense of potentially
introducing filter ringing.

FREQUENCY and GAIN

You can adjust an EQ band by clicking on a node and dragging it to change the frequency and gain
of the band. You can click and drag to select multiple filter nodes and move them as a group.
 EQ 158

Q/BANDWIDTH
If you move the mouse over the bracket handles on the side of the band, you can adjust the Q or
bandwidth of the EQ by dragging with the mouse. You can also adjust Q of the selected filter with
the mouse wheel.

More Information

Interacting with the EQ nodes

To adjust the EQ curves, grab an EQ node and drag it to a new point on the grid. When a node is
selected, handles on either side of it appear which can be dragged together or apart to control the
bandwidth of the node. Alternately, you can enter precise EQ settings by typing values into the table
below the main EQ grid.

Choosing between Analog and Digital EQ Types

There are reasons why you may want to use one EQ mode over another. Analog mode uses shapes
based on analog equalizers, implemented as digital IIR filters. One reason to use this mode is
that analog bells are narrower than our digital shapes at high Q values. Since these shapes are
minimum-phase, they cause no pre-echos. Also, since they match analog designs, they can be
used to emulate what an analog equalizer would do, or reduce damage caused by an analog
equalizer.
Digital mode uses special shapes designed to correct audio problems as surgically as possible,
implemented as FIR filters. These shapes, unlike the Analog shapes, affect only precisely defined
frequency ranges. In Digital mode the EQ is linear-phase, meaning no phase shift will occur.

EQ curve display & EQ Type

You may notice a slight difference in the visual shape of the bell filter when moving between
Analog and Digital modes. Analog mode follows the traditional shapes of analog circuitry, while
Digital mode uses special filters designed by iZotope. While these shapes are similar between the
two modes, they will not be identical.

Composite EQ Curve Display

As you adjust a band you will see two EQ curves. The white curve is the composite of all EQ bands
while the curves colored the same shade as the node shows the EQ curve of the selected band.
 EQ MATCH 159

48. EQ Match
ADV

Overview
The EQ Match module lets you match the EQ profile of a selection with the profile of a different
selection. This is useful if you’re ever tasked with matching a lav mic with a boom mic, matching
location dialogue to ADR or vice versa, or perhaps you had multiple mics on an audio source that
you’d like more closely aligned in terms of frequency response.

Controls

AMOUNT
Sets how closely the frequency spectrum you’re processing will be matched to the learned
spectrum. Many times, using a 100% match could sound unnatural, and lower values, between
10-40% do enough to make the two audio signals more closely aligned.

Workflow

How to Apply EQ Match to different selections

1. Open the EQ Match module
2. Make a selection in a file.
3. Click Learn.
4. Make another selection.
5. Click Process.
 EQ MATCH 160

How to save a captured spectrum as an EQ Match Preset

1. In EQ Match, click the preset menu to the right of the preset drop-down menu.
2. Select Add Preset.
3. Enter the name for the new preset.
4. Press Enter.
 FADE 161

49. Fade
Overview
The Fade module can be used to apply a gradual increase or decrease to gain throughout your
selection. Different amplitude curve options are available in the Fade Type menu. For in-phase
tonal material, a Cosine fade curve will work well, while the Equal power fade type can be more
effective on noisy material.

Controls

Fade Type
• Log: Logarithmic fade for general use
• Linear: Linear fade for general use
• Cosine: Cosine fade for crossfades of in-phase tonal audio material
• Equal power: Equal power fade for crossfades of noisy audio material

Instant Process Tool

STD & ADV
Fade is available as an option in the Instant Process menu. When Fade is selected and Instant
Process is enabled, the active settings in the Fade module will be applied to your selection. For
example: if the Fade module is set to Fade in: Log (as shown in the image above) the Fade in: Log
type of fade will be applied every time you use the Instant Process Tool in ‘Fade’ mode.
 GAIN 162

50. Gain
Overview
The gain module is useful for bringing the level of your audio up or down. Gain can also be applied
to a specific time-frequency selection, allowing you to manually attenuate or boost selections in
the Spectrogram window.

Controls

Gain
Boosts or cuts the level of the signal by the designated decibel amount.

Instant Process Tool

STD & ADV
Gain is available as an option in the Instant Process menu. When Gain is selected and Instant
Process is enabled, the active settings in the Gain module will be applied to your selection.
 LEVELER 163

51. Leveler
ADV

Overview
The Leveler module automatically rides the gain in your file to even out the variations of the signal
level. The algorithm consists of a compressor with a makeup gain to achieve a smooth signal
that’s aiming towards (though may not exactly hit) a desired Target RMS level. The compressor
has the ability to prevent pumping on speech pauses or breathing sounds, using the Optimization
mode, for either Dialogue or Music, in addition to the Ess and Breath parameters.
The level detector stage includes the K-weighting filter that helps equalize the audible loudness, not
just RMS level. However, the Leveler module is designed for the smoothing of overall audio signals,
rather than taking an entire signal and using a fixed gain to ensure it hits a loudness compliant
LKFS level, which is the goal of the Loudness module.
This all combines to create a transparent, non-destructive Clip Gain curve, without the color or
artifacts of a traditional compressor.
Unlike the Loudness module, which applies a constant gain based on Loudness compliant analysis
to the whole file, Leveler applies a time-variable gain. For convenience of RX users, the time-
variable gain is applied as a Clip Gain envelope, which can be viewed and edited by the user.

PLEASE NOTE
If additional processing is applied from another module after running Leveler, the clip
gain values assigned by Leveler will be destructively written to the file and the clip gain
nodes will return to zero. However, the clip gain settings from Leveler will be saved in
the Undo History list).
 LEVELER 164

Controls

Numerical Readouts
Numerical Readouts provide you with the Total, Maximum and Minimum readouts for RMS. The
total value is the overall RMS of your audio signal, which may inform where you choose to set the
Target RMS level parameter.

Optimize For
Optimize For switches between two modes, Dialogue and Music. Each mode utilizes a slightly
different handling of the noise floor.
• Dialogue tends to be audibly juxtaposed against the noise floor, as it’s typically very transient,
whereas music often tends to fade into the noise floor, with chords, notes, and other
instrumental decays.
• Switching between these two modes will affect the behavior of the Leveler, and prevent
pumping.

Target Level
Target Level sets the desired average RMS level of the recording.
• Note that Leveler uses K-weighted RMS to better level perceived loudness, but that it is not
a loudness compliant leveling tool. It uses the Target Level as a guide, but with the goal
of smoothing out variations in an audio signal much more transparently than a compressor
typically would. As such, it is not unusual to see the resulting output of Leveler not be an exact
1:1 with the defined Target Level.
• At high target levels, the leveler may not be able to hit the target without clipping, so the target
level will not be reached.
 LEVELER 165

Responsiveness
Sets the integration time for RMS level detection and is similar to the attack/release setting on a
compressor.
• Lower settings will result in more aggressive Leveling, useful if a signal has a lot of sudden
variations.
• Higher settings will result in smoother behavior, leveling words or phrases rather than individual
syllables.
• If you find the Leveler is responding to any sudden unwanted sounds, such as a cough, and
boosting it, increase the slider to a higher value to see if this results in less aggressive jumps.

Preserve Dynamics
This can be thought of the maximal amount of gain applied by the Leveler. The wider the range of
gain adjustments allowed, the further away from the original dynamic range the audio signal will
be.
• At lower values, the Leveler will preserve fewer of the original dynamics in the audio signal.
• At higher values, the Leveler will preserve more of the original dynamics in the audio signal.

Ess Reduction
Ess Reduction is aimed at anyone using the Leveler on dialogue or vocals, and utilizes a smart
algorithm, inspired by the DBX 902 De-esser, to detect when ess is present in a signal, and then
attenuate it accordingly. This avoids adding any boost to esses, which may otherwise be seen as
quiet sounds requiring a boost. The slider sets the amount of ess reduction, applied in dB.

Breath Control
Breath Control will automatically detect breaths in your vocal takes and attenuate them. This saves
time when editing dialogue or vocal tracks, and streamlines a task that is typically done manually.
• Breath Control automatically analyzes the incoming audio and distinguishes breaths based on
their harmonic structure. If any piece of the incoming audio matches a harmonic profile similar
to a breath, the Leveler will apply a Clip Gain adjustment.
• Different from a ‘Threshold’ based process in which the module is only engaged once the audio
has risen to a certain volume, this feature will perform its analysis regardless of level.
• This allows for accurate breath recognition with a multitude of quiet or loud dialogue / vocal
styles with minimal adjustment of the module’s controls.
• The slider represents the desired level, in dB, that you wish all detected breaths to be reduced
to. This can result in much more natural sounding breath reduction as the detected breaths in
your audio are only reduced when necessary.
• Loud and abrasive breaths will be reduced heavily, and quiet, natural sounding breaths will be
left at the same volume. The volume level specified by this slider is a guide, but may not result
in exact values.

Limiter
The Leveler has a built-in Limiter in order to avoid introducing any clipping to the audio signal
once the Clip Gain envelope has been applied. This cannot be adjusted, but you’ll see the Clip Gain
envelope smooth off an audio signal if you’re pushing peaks close to 0 dB.
 LOUDNESS CONTROL 166

52. Loudness Control

Overview
The Loudness Control module adjusts the gain of the signal in order to meet the specified
loudness standard, such as BS.1770. Unlike the Leveler module, the gain in the Loudness Control
module stays fixed in time. However a post-limiter will be applied to the signal, if it is required to
meet the True peak specification.

Controls

Integrated loudness
Sets the desired integrated loudness of the clip, in LKFS (which is the same as LUFS).

Tolerance
Sets the +/- margin of error for the integrated loudness, in LU (loudness units). Each standard
determines the tolerance, which is typically set between 0.5 - 2.

Short-term/Momentary Loudness
Short term loudness is calculated over a moving window of 3 seconds. Some loudness standards
like EBU R128 s1 Short Term define a maximum short-term value. Momentary loudness is
calculated over a moving window of 400ms. The Short-term/Momentary power button enables
or disables this processing, and the drop-down allows you to choose between short term or
momentary loudness.

True peak
Sets the maximum true peak value allowed in a clip.
 LOUDNESS CONTROL 167

Program loudness gate

When computing Integrated loudness, certain low-level signals are excluded from the computation
in order to only measure the loudness of “typical” signal levels. This is called gating and is a part of
the BS.1770-2, 3 standards (and also R128). Firstly, all signals with a Momentary loudness below
−70 LKFS are excluded, so that the noise floor in pauses does not bias the computed loudness.
Secondly, all signals with a lower Momentary loudness than 10 dB (or LU) below the average
Momentary loudness are also excluded. This allows the computation of the Integrated loudness
only from the typical and loudest parts of the program, while excluding silence and quieter parts,
which have less effect on the listener’s impression of overall loudness of the program . Most of the
loudness standards shipping with RX Loudness Control use gating, an exception being BS.1770-1.
This parameter allows you to toggle gating on or off, if you’re working to a unique requirement.

Workflow
To use the Loudness Control module:
1. Make a selection in your file. If you want to apply Loudness Control to the entire clip, use
ctrl/cmd + A or click Edit > Select All.
2. Select a preset, or make adjustments to the controls manually.
3. Click Render.

Loudness Standards
Loudness Integrated Integrated Short- Momentary True Gating Notes
Standard Tolerance term Peak
(+/-)
AES AGOTTVS -16 2.0 Off Off -1 On
TD1006.1.17-10
AGCOM -24 0.5 Off Off -2 On Italy
219/09/CSP
ARIB TR-B32 -24 2 Off Off -1 On Japan
A/85
ATSC A/85 -24 2 Off Off -2 On USA,
Canada,
Puerto Rico
BS.1770-1 -24 2 Off Off -2 Off International
BS.1770-2/3/4 -24 2 Off Off -2 On International
EBU R128 -23 0.5 Off Off -1 On Europe
EBU R128 DPP -23 0.5 Off Off -3 On Europe
EBU R128 -23 1 Off Off -2 On South Africa
(South Africa)
OP-59 -24 1 Off Off -2 On Australia/Ne
w Zealand
Portaria 354 -23 0.5 Off Off -2 On Brazil
 MIXING 168

53. Mixing
Overview
Provides specific control over both left and right signal and balance levels. This simple operation
can be used to downmix stereo material into mono, invert waveforms, transcode left/right stereo
into mid/side, subtract a center channel, and much more.

Controls

Left Output Mix (%)

Allows you to define how much of the current selection’s left and right channel signal will be
present in the new target left channel.

Right Output Mix (%)

Allows you to define how much of the current selection’s left and right signal will be present in the
new target right channel.
 NORMALIZE 169

54. Normalize
Overview
The Normalize module applies enough gain to set the sample peak level of your signal to the
specified Target Peak Level.

Target Peak Level

Target Peak Level [dBFS]: Determines the maximum peak level of a signal as a result of
normalization.
 PHASE 170

55. Phase
Overview
The Phase module balances asymmetric waveforms by rotating signal phase. Rotating the phase
of a signal changes its peak values but doesn’t change its loudness, and otherwise has no audible
effect on the signal.

Controls

Adaptive Phase Rotation

Continuously analyzes the audio selection and applies the time-variable phase rotation to both left
and right channels, resulting in a symmetrical waveform with minimal signal peak levels.
Adaptive phase rotation is best used on vocal material, as it can occasionally yield pitch artifacts
on musical material.

Rotation (deg)
Rotates the channel’s phase by the specified degree.
When a waveform’s phase is rotated, every frequency is rotated equally. Rotating phase by 180
degrees inverts the waveform.

Suggest
Analyzes the selection for the ideal channel-linked fixed phase for reducing overall peak levels of
the signal.

More Information
Asymmetric waveforms can occasionally occur in audio such as dialogue, voice, and brass
instruments.
• Making the waveform more symmetrical gives the signal more headroom.
• Rotating the phase of a waveform, will change its amplitude characteristics. Phase rotation
does not result in a time shift.
 PHASE 171

• Because the range of rotation is from −180 to +180 degrees, the Phase tool can be used for
simpler purposes, such as inverting signal polarity.

Visual Example of Phase Rotation

The top waveform in the following image is a trumpet signal with higher peak values on one
side of its waveform (meaning the waveform is asymmetrical.) The bottom waveform in the
following image has been processed by the Phase module. The processig rotated the phase of the
waveform by −72 degrees to distribute its peak samples more evenly (making the waveform more
symmetrical.)
 PLUG-IN HOSTING 172

56. Plug-in Hosting

Overview
On Windows, RX supports VST3 and VST2 plug-ins. On Intel-based macs and Apple silicon-based
macs running RX with Rosetta, RX supports AU, VST3, and VST2 plug-ins. For macs running RX
natively on Apple silicon, AU and VST3 are supported. VST2 is not supported when RX is running
natively on Apple silicon.

UPDATE YOUR PLUG-INS

Please make sure your third-party plug-ins are updated to their latest version in
order to ensure that RX will be able to scan them correctly. Contact your plug-in
manufacturers for updated installers, if necessary.

Selection Based Processing

With a plug-in loaded in the Plug-in module, you can make use of the same audio selection tools
and Preview and Compare options that are available for other RX modules.
This can allow for very detailed processing and greater accuracy when working with your existing
plug-ins, giving you audio selection options unavailable in a traditional DAW setup.

Plug-in Presets
Only one plug-in may be loaded at a time. However, with the use of presets, multiple settings and
presets may be recalled quickly in order to move between plug-in instances. When your plug-in is
configured the way you would like it, select Add Preset from the small Preset drop down arrow.
 PLUG-IN HOSTING 173

Assign Presets to Keyboard Shortcuts

Once your preset is named and saved, you can then assign that plug-in and preset to a keyboard
shortcut with the Set Preset Shortcut feature.
This keyboard shortcut will recall not only the plug-in settings, but the plug-in instance itself. As
such, if a preset is saved with Plug-in 1, and Plug-in 2 is currently loaded into RX’s plug-in window,
pressing the preset keyboard shortcut will re-instantiate Plug-in 1 and recall the exact settings
when the preset was made.
This can allow for very quick editing, processing, and recall of plug-in instances and settings,
providing a quicker workflow than traditional DAW track/mixer environments.
 RESAMPLE 174

57. Resample
Module Only

Overview
The Resample module allows you to convert an audio file from one sampling rate to another.
Sample Rate Conversion (SRC) is a necessary process when converting material from one
sampling rate (such as studio-quality 96 kHz or 192 kHz) to another rate (such as 44.1 kHz for
CD or 48 kHz for video).
It is common to record and edit in high sampling rates since higher rates allow higher frequencies
to be represented. For example, a 192 kHz audio sample can represent frequencies up to 96 kHz
whereas a 44.1 kHz audio sample can only represent frequencies up to 22.05 kHz. The highest
frequency that can be represented accurately by a sampling rate is half of the sampling rate, and is
known as the Nyquist frequency.
When reducing the sampling rate, or downsampling, it is crucial to remove the frequencies that
cannot be represented at the lower sampling rate. Leaving frequencies above this point causes
aliasing. Aliasing can be heard as the frequencies in an inaudible range are shifted into an audible
range, causing distortion and noise. With iZotope SRC’s steep low-pass filter, users can completely
avoid the common aliasing artifacts while maintaining the maximum frequency content.
Comparison of iZotope's SRC process
A comparison of iZotope’s SRC process versus other sample rate convertors can be viewed at:
http://src.infinitewave.ca/.

Controls

NOTE ABOUT THE RED ALIASING CURVE DISPLAY

The Aliasing portion of the curve displayed in red shows the reflected frequencies
during downsampling or imaged frequencies during upsampling — both due to
aliasing.
 RESAMPLE 175

New sampling rate

This setting chooses the sampling rate you want to convert to. Choose a sampling rate from the
drop-down list, or click on the field to type in a custom sampling rate.

Change tag only

Changes the declared sampling rate of the file in the file’s properties without resampling the file,
effectively changing the playback rate and pitch of the file.

WHEN TO USE CHANGE TAG ONLY OPTION

This feature is useful if the sampling rate tag was damaged by a previous audio
editing process and the file is playing back incorrectly.

Filter steepness
This allows you to control the steepness of the SRC filter cutoff. The white line is representative of
an ideal low-pass filter.

NOTE ABOUT HIGHER FILTER STEEPNESS VALUES

Higher filter steepness means better frequency performance of the filter: wider
passband retains more useful signal, while stronger stopband attenuation provides
better rejection of aliasing. At the same time, higher steepness of the frequency
response requires a longer filter, which produces more ringing in time domain and
energy smearing near the cutoff frequency.

Cutoff shift
SRC filter cutoff frequency shift (scaling multiplier). Allows shifting the filter cutoff frequency up or
down, to balance the width of a passband vs. amount of aliasing.

Pre-ringing
SRC filter pre-ringing amount in time domain (0 for minimum phase, 1 for linear phase, or
anywhere in between). Adjusts the phase response of the filter, which affects its time-domain
ringing characteristic. The value of 0 produces a minimum-phase filter, which has no pre-ringing,
but maximal post-ringing. The value of 1 produces a linear-phase filter with a symmetric impulse
response: the amount of pre-ringing is equal to the amount of post-ringing. Intermediate values
between 0 and 1 produce so-called intermediate-phase filters that balance pre- and post-ringing
while maintaining linear-phase response across a possibly wider range of frequencies.

Post-limiter
Keeps true peak levels of the output signal below 0 dBTP to prevent any clipping from occurring.
When to use the post-limiter:
• This option is important when resampling signals that are very close to 0 dB, because filtering
during resampling can change peak levels of a signal.
 RESAMPLE 176

• Engage the Post-limiter option in order to limit the output levels of your signal to prevent any
clipping from occurring.
 SIGNAL GENERATOR 177

58. Signal Generator

Overview
The Signal Generator is able to synthesize silence, tones, and noise. This is useful for creating test
tones, calibration tones for post production delivery specs, repairing DC offset, and even using it as
a “bleep” module to eliminate obscenities in a dialogue edit.
The Signal Generator module is capable of generating extremely accurate test tones for research
and testing purposes.

Silence
Silence generates digital silence, which can be used to adjust spacing or duration of your file.

DC Offset
Creates or removes positive or negative amounts of DC offset.

Tones
The Tones tab allows you to generate signals.

• TONE SHAPE: Tone shape allows you to choose the waveform for the signal to be generated,
with a choice of: Sine, Triangle, Sawtooth, Square.
 SIGNAL GENERATOR 178

• FREQ: Sets the frequency, in Hertz (Hz), of the signal to be generated.

• AMP: Sets the amplitude, in Decibels (dB), of the signal to be generated.
• ANTIALIASING: All test tone shapes, except the sinusoid, contain an infinite set of harmonics.
When Antialiasing is not selected, these tones are generated in a “naive” way in the time
domain. It leads to aliasing (folding) of higher harmonics into the audible range, which often
sounds undesirable.
• When Antialiasing option is enabled, these higher harmonics are filtered out with a linear-
phase low-pass filter. It prevents aliasing, but introduces some ringing into the waveform,
which is usually not a problem, because it is concentrated near half the sampling rate.
• Higher quality of antialiasing produces a filter with a sharper frequency cut, which retains
more of the useful harmonics and rejects more of the aliased harmonics, at the expense of
more time-domain ringing and slower processing.
• LEVEL: Specifies the sample peak level of the synthesized waveform in dBFS. When
Antialiasing is used, actual sample and true peak levels may exceed the specified peak level due
to filter ringing.

Noise

Allows you to generate noise of many different types, with a user definable RMS level. The noise
types include: White Gaussian, White triangular, White uniform, White binary, Pink, Brown.
Noise color defines the spectral shape: white noise has flat power spectrum, pink noise decays at 3
dB/octave rate, while brown noise decays at 6 dB/octave rate.
Gaussian, triangular, uniform, or binary relates to the p.d.f., or probability distribution function, of
the noise. It describes how often samples of different amplitude are encountered in the signal. For
example, uniform p.d.f. means that all amplitudes below noise peak level are equally popular.
• PASTING MODE: The pasting modes affect how generated signals are added to your file.
• REPLACE: Replace will completely replace the audio in your selection with the generated
signal.
• Mix: Mix will retain the audio in your selection, and mix in the generated signal.
• Insert: Insert will allow you to set a custom duration, in seconds, and the Signal Generator
will then insert this audio, increasing the length of your audio file by the amount prescribed.
 TIME & PITCH 179

59. Time & Pitch

Overview
The Time & Pitch module uses the iZotope Radius™ processing algorithm to allow for independent
control over the length and pitch of your audio. It is useful for retuning audio to fit in a mix better, or
adjusting the length of a selection to account for changes in BPM or timecode.
• The Variable Pitch module can be used for faster pitch shifting with the ability to correct
variations in pitch over time.
• The Variable Time module can be used to adjust the time stretch ratio envelope over time.

iZotope Radius
iZotope Radius™ is a world-class time-stretching and pitch-shifting algorithm. You can easily
change the pitch of a single instrument, voice, or entire ensemble while preserving the timing
and acoustic space of the original recording. iZotope Radius is designed to preserve the natural
timbral qualities of the original file, even when applying extreme pitch shifts.

Controls

Algorithm
The Algorithm drop-down menu has three options:

Radius
Designed to work well with polyphonic material such as mixes with more than one instrument, as
well as non-harmonic material such as drum loops or rhythmic audio. This is the highest-quality
option for most sources.

Radius RT
Good quality, polyphonic, but faster than Radius. If processing speed is important, use the Radius
RT algorithm.
 TIME & PITCH 180

Solo Instrument
Designed for monophonic pitched material such as a stringed instrument or human voice. You
should use Solo mode only when processing a single instrument with a clearly defined pitch.
The human voice is a good candidate for solo mode, as are most stringed instruments, brass
instruments, and woodwinds.
In Solo mode, the adaptive window size can significantly affect the quality of Radius’s output.
• If the adaptive window size is too small, you will hear a squeaking noise which sounds like the
pitch of the audio is changing very rapidly.
• If the adaptive window size is too large then the sound will become grainy as you will begin to
hear portions of it being repeated.
• A good approach is to start with the default window size of 37 ms. If the results are
unsatisfactory, increase the window size until the squeaking noise described above does not
occur. If you cannot get the distortion to disappear, switch to Radius mode for processing.
• Lower pitched instruments and voices may require a longer adaptive window size than the
default, but very long adaptive window sizes can cause audible repeating slices of audio.

Formant Correction
Formants are the resonant frequency components of voice that tend to be perceived as
characteristics like age and gender. You can shift formants independently of pitch and time by
enabling Shift Formants.
• Typically you will leave the Formant Shift Strength set to 1 (full strength) and the Formant Shift
Semitones set to 0.
• If you hear what sounds like an EQ adjustment to your audio, you can try lowering the strength
to reduce this artifact.
• To achieve special effects, for example to change the perceived gender of a human voice, try
adjusting the semitones to a value other than 0.

Stretch Ratio
Determines how much the resulting audio will be stretched in time.
• Values between 12.5% and 100% will cause the audio to speed up without affecting pitch,
resulting in a shorter audio file.
• Values between 100% and 800% will cause the audio to slow down without affecting pitch,
giving you a longer audio file.

BPM Calculator
If you are using Radius to process audio for a tempo change, you can also adjust the stretch ratio
with the BPM Calculator.

Pitch Shift
Controls the amount of pitch shifting up or down that will be applied to the audio.

Transient Sensitivity
Determines the algorithm’s handling of transient material. Higher values will result in better
preservation of individual transients after processing.
 TIME & PITCH 181

• When stretching percussive material, you usually want transient sensitivity set to its default
value of 1.
• If transients in your audio are being “smeared”, a higher value of 2 will tighten up transience at
the expense of incurring heavier processing on non-transient audio.
• Bowed instruments such as the violin and cello are especially affected by the transient
sensitivity setting. If you hear a stuttering artifact, lower the transient sensitivity to eliminate
it.

Noise Generation (Radius Mode Only)

Helps noisy material (like sibilance or snare drums) sound more natural when processed.

NOTE
This control will generate noise instead of stretching the noise that is already present
in the signal and creating new tones. Higher values of the noise generation parameter
will cause Radius to generate noise more often, but can cause some phase artifacts.

Pitch Coherence (Radius Mode Only)

The Pitch coherence control in the Radius control panel helps preserve the natural timbre for
pitched solo voices, such as human speech, saxophone or vocals. While traditional vocoders
can smear these signals in time and randomize phase, the pitch coherence parameter of Radius
preserves phase coherence for these signals.

High values of pitch coherence will avoid phasiness in Radius’s output at the expense
of roughness (modulation) in processed polyphonic recordings. Try turning this up for
better results if you’re processing a solo voice or a small group of related instruments.

Phase Coherence (Mix Mode Only)

Preserves the coherence of phase between the left and right channels of the processed audio.
This should be increased if there’s any change in the perceived stereo image after using Radius.
It can be decreased when processing a multichannel signal where different channels contain
completely different instruments.

Adaptive Window Size (ms) (Solo Mode Only)

Adjusts the window size in milliseconds of Radius’ Solo algorithm.
• If the adaptive window size is too small, you will hear a squeaking noise which sounds like the
pitch of the audio is changing very rapidly.
• If the adaptive window size is too large then the sound will become grainy as you will begin to
hear portions of it being repeated.
• Increase this if you have trouble getting good results pitching or stretching low-pitched
instruments or voices.
 TIME & PITCH 182

Shift Formants
Processes formant frequencies independently of other pitch and time processing.
• When this option is enabled, formant frequencies can be shifted independently of other pitch
shifting performed by Radius.
• When Radius performs pitch-shifting without Formant Correction, it will shift these resonant
frequencies along with the rest of the audio.

Strength
Adjusts the amplitude strength of the formant correction filter.

Shift
How much formant frequencies are shifted. Typically this control can be set to 0, which leaves
the formant frequencies unshifted during processing. Adjust this control to fine-tune the formant
correction algorithm or for special effects.

Width
Controls the bandwidth of the formant detection filter.
• Smaller values of this control will offer more precise formant correction in the processed audio.
• Higher values will include a wider band of formant frequencies.

More Information
Pitch shifting single instruments (especially bass instruments) can benefit from some adjustments
to formant correction. Try enabling formant correction and moving the strength between 0.1 and
0.2. Move the Formant Correction semitones part of the way towards your pitch shift amount. For
example, if you’re pitch shifting +4 semitones, move the Formant Correction Semitones between 2
and 3. This can help bring back subtle percussive elements in the original source material.
The formant frequencies of the human voice can actually shift slightly when we sing. You can use
the Formant Correction Semitones control to compensate for this. For example, if pitch shifting a
human voice by +7 semitones, try setting the Formant Correction semitones between 0 and +2 for
more natural results.
 VARIABLE PITCH 183

60. Variable Pitch

Overview
Variable Pitch can be used to quickly fix small pitch variations or to correct gradual pitch drifts
over time. With the addition of Preserve Time mode, Variable Pitch replaces and expands on the
processing options available in the Pitch Contour module available in previous versions of RX.

Displays
Variable Pitch features a waveform panel and a spectrogram panel that each display information
about the current selection in the active file tab. These panels will dynamically update when the
selection is changed. If no selection is made in the active file tab, no information will be displayed
in the spectrogram or waveform panels.

WINDOW RESIZING
Click and drag on the bottom right-hand corner of the module window to customize
the window size.

Waveform Display
The single waveform drawn in this panel represents the sum of all enabled channels in the current
selection. The waveform drawing is normalized to allow for consistent vertical resolution when
working with selections of varying amplitude.
 VARIABLE PITCH 184

Spectrogram Display
The spectrogram drawn in this panel represents the sum of all enabled channels in the current
selection.

Playhead Indicators
The solid white vertical line and dotted yellow vertical line overlaid on the waveform and
spectrogram panels indicate the current playhead position (white) and the playhead anchor
position (yellow).

Current Playhead Position

The solid white vertical line overlaid on the waveform and spectrogram panels indicates the
current playhead position. This indicator line updates to follow the current playhead position during
playback. The playhead position indicator will only appear in the module window when it is within
the bounds of the current selection.

Playhead Anchor
The dotted yellow vertical line overlaid on the waveform and spectrogram panels indicates the
playhead anchor position in the main editor window. If the playhead anchor position is outside of
the current selection bounds, the indicator will not be displayed in the module window.
 VARIABLE PITCH 185

Contour Curve Display

The blue line overlaid on the spectrogram panel represents the pitch contour curve. Nodes can be
added to this curve and adjusted to make changes to pitch over the course of the active selection.

Contour Curve Axes

The contour curve allows for adjustments along two axes: Pitch and Time.
• Pitch: The vertical y-axis of the contour curve represents pitch in semitones.
• The Pitch axis ranges from -24 (bottom) to +24 (top) semitones.
• The center of the Pitch axis equates to 0 semitones.
• Time: The horizontal x-axis represents time.
• The time format used here is determined by the Time Format Display selection in the
transport section of the main editor window.
• The range of the time ruler matches the length of the current selection.
To zoom the ruler in and out:
1. Hover over the ruler and use a mousewheel or trackpad to zoom in and out.
2. Click and drag left or right on the ruler when zoomed in to change the ruler position.
3. Double-click on a ruler display to reset the zoom level to default.

Contour Curve Readout

When the cursor is positioned over the spectrogram panel, a text readout will appear in the upper
left hand corner of the panel. This readout displays information about the processing that will be
applied by the current contour curve.

The readout displays the following information about the current cursor position, from left to right:
• Time: Current time position of the cursor within the spectrogram panel.
• Pitch Shift (%): Percentage of pitch shift that will be applied at the cursor’s current time
position.
• Pitch Shift (Semitones): Amount of pitch shift that will be applied at the cursor’s current time
position.

Contour Curve Editing

The following section describes the methods and controls available for editing the contour curve.
 VARIABLE PITCH 186

Add Nodes
Click in the spectrogram panel to add a new node to the contour curve.

CONTOUR CURVE NODE LIMIT

The contour curve supports adding up to 25 nodes.

Semitone Adjustments
Click and drag a node up or down to adjust its semitone value.

Time Adjustments
Click and drag a node left or right to move the associated pitch adjustment point earlier or later in
time.
• Nodes cannot be moved outside of the time bounds of the current selection.
• The contour curve shape will be maintained when the selection changes.
• The contour curve shape will be maintained after rendering.

Remove Nodes
Individual nodes can be deleted from the curve using the following methods:
▶ Click and drag a node past the top or bottom border of the contour curve display to quickly
remove it from the curve.
▶ Control-click (Mac) or ctrl-click (Windows) on a node to remove it from the curve.

Reset Individual Nodes

Double-click on a node to reset it to the default value of 0 semitones. Double-clicking a node only
resets the semitone value to default, it will not change the time position of the node.

Reset Curve
Removes all custom nodes from the curve, resetting it to default. Two nodes are present in the
default curve, one at the start and one at the end of the current selection. The default nodes are set
to 0 semitones (no pitch adjustment).

Smoothing
Adjusts the amount of smoothing applied between nodes on the contour curve. Smoothing is a
global control and is applied to all nodes on the curve.
• Lower smoothing values: Applies little to no smoothing between nodes on the curve, resulting
in strict transitions between nodes.
 VARIABLE PITCH 187

• Higher smoothing values: Applies more smoothing between nodes on the curve, resulting in a
gradual, rounded slope between nodes.

Controls
The following section describes the Preserve Time control and the parameters available for
refining the results of processing when Preserve Time (Radius Mode) is enabled.

Preserve Time
Determines the pitch processing algorithm that is used when rendering the current selection.
When Preserve Time is enabled, the iZotope Radius™ processing algorithm is used. When Preserve
Time is disabled, a resampling processing algorithm is used. See the descriptions below for more
information about these processing algorithms and how they affect the rendered output.

Preserve Time Disabled (Resampling Mode)

When Preserve Time is disabled, a resampling algorithm is used to render the pitch contour curve.
In this mode, time and pitch are adjusted synchronously, as if tape playback speed was adjusted.
The length of the rendered selection will change to account for changes in pitch.
• Negative semitone adjustments increase the length of the rendered selection.
• Positive semitone adjustments decrease the length of the rendered selection.

USE CASE FOR RESAMPLING MODE

This mode can be useful for synchronizing two recordings of the same performance
that have been captured at slightly different clock speeds.

Preserve Time Enabled (Radius Mode)

When Preserve Time is enabled, the Radius algorithm is used to render the pitch contour curve.
The mode will adjust pitch without altering the length of the current selection.

PRESERVE TIME: ADDITIONAL CONTROLS

When Preserve Time mode is enabled, additional controls become available for
refining the results of processing: Pitch Coherence & Transient Sensitivity.

Pitch Coherence
Adjusts the amount of timbre preservation applied during processing. This control is useful for
maintaining the natural timbral qualities of pitched audio content (such as human speech, sung
vocals or saxophone) by correcting the smearing or phasing issues that can be introduced during
processing.
 VARIABLE PITCH 188

Increasing pitch coherence can help to combat phase effects that can occur when applying
processing to a solo voice or a small group of related instruments. A possible tradeoff of
increasing pitch coherence is that it may introduce roughness or unwanted modulation when
applying processing to polyphonic or multi-instrument content.

NOTE
Pitch Coherence is only available for adjustment when Preserve Time is enabled.

Transient Sensitivity
Adjusts how the Radius algorithm detects and subsequently preserves transient content when
processing. Higher values can improve transient clarity at the cost of applying heavier processing
to non-transient material. Bowed instruments, such as cello or violin, may present stuttering
artifacts after processing with higher transient sensitivity values. Lowering the transient sensitivity
value can help to reduce the unwanted artifacts in sustained content.

TRANSIENT SENSITIVITY CONTROL AVAILABILITY

Transient Sensitivity is only available for adjustment when Preserve Time mode is ON.

Compare Settings
The Variable Pitch module does not allow for real-time preview playback. To audition different
settings before rendering, click the Compare button in the module footer area to send the settings
to the Compare Settings window.

Learn more about Compare Settings in the Common Module Controls.

Alternative Modules
• For pitch envelope editing tailored to speech, try using the Dialogue Contour module in RX 11
Advanced.
• Make static time-stretch and pitch-shift adjustments with the Time & Pitch module.
• Make adjustments to the time stretch ratio envelope of a selection with the Variable Time
module.
 VARIABLE TIME 189

61. Variable Time

Overview
Variable Time can be used to correct timing issues or creatively adjust speed over the course
of a selection. This module leverages the iZotope Radius algorithm to allow for high-quality time
compression and expansion without also adjusting pitch.

Displays
Variable Time features a waveform panel and a spectrogram panel that each display information
about the current selection in the active file tab. These panels will dynamically update when the
selection is changed. If no selection is made in the active file tab, no information will be displayed
in the spectrogram or waveform panels.

WINDOW RESIZING
Click and drag on the bottom right-hand corner of the module window to customize
the window size.

Waveform Display
The single waveform drawn in this panel represents the sum of all enabled channels in the current
selection. The waveform drawing is normalized to allow for consistent vertical resolution when
working with selections of varying amplitude.
 VARIABLE TIME 190

Spectrogram Display
The spectrogram drawn in this panel represents the sum of all enabled channels in the current
selection.

Contour Curve Display

The blue line overlaid on the spectrogram panel represents the time stretch ratio contour curve.
Nodes can be added to this curve and adjusted to make changes to time stretch ratio over the
course of the active selection.

Contour Curve Axes

The contour curve allows for adjustments along two axes: Time Stretch Ratio (y-axis) and Time
(x-axis).
• Time Stretch Ratio: The vertical y-axis of the contour curve represents time stretch ratio values
in increments associated with both Speed (multiplier) and Length (percentage) adjustments.
The effect of adjustments along this axis can be thought of as affecting the processed signal in
two ways, adjusting length or adjusting speed. The minimum and maximum values of this axis
equate to the following speed and length adjustments:
• Minimum value:
• Speed: 0.125x - 0.125 times slower than the original
• Length: 800.0% - 8 times longer than the original
• Maximum value:
• Speed: 8.000x - 8 times faster than the original
• Length: 12.5% - 1/8th the length of the original; 8 times shorter.
• The center of the y-axis is equivalent to no adjustment in length or speed.
• Time (Current Selection Length): The horizontal x-axis represents time across the current
selection.
• The time format used here is determined by the Time Format Display selection in the
transport section of the main editor window.
• The range of the time ruler matches the length of the current selection.
 VARIABLE TIME 191

To zoom the ruler in and out:

1. Hover over the ruler and use a mousewheel or trackpad to zoom in and out.
2. Click and drag left or right on the ruler when zoomed in to change the ruler position.
3. Double-click on a ruler display to reset the zoom level to default.

Contour Curve Readout

When the cursor is positioned over the spectrogram panel, a text readout will appear in the upper
left hand corner of the panel. This readout displays information about the processing that will be
applied by the current contour curve when it is rendered.

The readout displays the following information about the current cursor position, from left to right:
• Time: Current time position of the cursor within the spectrogram panel.
• Length Adjustment (%): Percentage of length adjustment that will be applied at the cursor’s
current time position.
• Speed Adjustment (x - multiplier): Amount of speed adjustment that will be applied at the
cursor’s current time position.

Contour Curve Editing

The following section describes the methods and controls available for editing the contour curve.

Add Nodes
Click in the spectrogram panel to add a new node to the contour curve.

CONTOUR CURVE NODE LIMIT

The contour curve supports adding up to 25 nodes.

Time Stretch Ratio Adjustments

Click and drag a node up or down to adjust its time stretch ratio value.

Time Adjustments
Click and drag a node left or right to move the associated time stretch ratio adjustment earlier or
later in time.
• Nodes cannot be moved outside of the time bounds of the current selection.
• The contour curve shape will be maintained when the selection changes.
 VARIABLE TIME 192

• The contour curve shape will be maintained after rendering.

Remove Nodes
Individual nodes can be deleted from the curve using the following methods:
▶ Click and drag a node past the top or bottom border of the contour curve display to quickly
remove it from the curve.
▶ Control-click (Mac) or ctrl-click (Windows) on a node to remove it from the curve.

Reset Individual Nodes

Double-click on a node to reset it to the default value of 100%/1.000x (no change to time stretch
ratio). Double-clicking on a node will only reset its time stretch ratio value to default, it will not
change the time position of the node within the selection.

Reset Curve
Removes all custom nodes from the curve, resetting it to default. Two nodes are present in the
default curve, one at the start and one at the end of the current selection. The default nodes are set
to 100%/1.000x (no time stretch ratio adjustment).

Smoothing
Adjusts the amount of smoothing applied between nodes on the contour curve. Smoothing is a
global control and is applied to all nodes on the curve.
• Lower smoothing values: Applies little to no smoothing between nodes on the curve, resulting
in strict transitions between nodes.

• Higher smoothing values: Applies more smoothing between nodes on the curve, resulting in a
gradual, rounded slope between nodes.

Controls
The following section describes the controls available for refining the Radius algorithm processing
results.

Pitch Coherence
Adjusts the amount of timbre preservation applied during processing. This control is useful for
maintaining the natural timbral qualities of pitched audio content (such as human speech, sung
vocals or saxophone) by correcting the smearing or phasing issues that can be introduced during
processing.
 VARIABLE TIME 193

Increasing pitch coherence can help to combat phase effects that can occur when applying
processing to a solo voice or a small group of related instruments. A possible tradeoff of
increasing this control is that it may introduce roughness or unwanted modulation when
processing polyphonic or multi-instrument content.

Transient Sensitivity
Adjusts how the Radius algorithm detects and subsequently preserves transient content when
processing. Higher values can improve transient clarity at the cost of applying heavier processing
to non-transient material. Bowed instruments, such as cello or violin, may present stuttering
artifacts after processing with higher transient sensitivity values. Lowering the transient sensitivity
value can help to reduce the unwanted artifacts in sustained content.

Compare Settings
The Variable Time module does not allow for real-time preview playback. To audition different
settings before rendering, click the Compare button in the module footer area to send the settings
to the Compare Settings window.

Learn more about Compare Settings in the Common Module Controls chapter.

Alternative Modules
• Make static time-stretch and pitch-shift adjustments with the Time & Pitch module.
• Adjust the pitch envelope over the course of a selection with the Variable Pitch module.
• For pitch envelope editing tailored to speech, try using the Dialogue Contour module in RX 11
Advanced.
 FIND SIMILAR 194

62. Find Similar

Overview
There are times when you may have many similar audio events to manually repair. In cases like
this, manually selecting and processing each event can be time consuming. RX includes a Find
Similar Event tool which takes your selection and finds all related instances of that audio event.
You can choose from Find Next, Find Previous, and Find All.

Controls

• Similarity: Lower values will find more events. The higher the value, the more similar an event
must be to the original event selection for it to be detected.
• Find Previous: Searches before the current selection for a similar event.
• Find Next: Searches after the current selection for a similar event.
• Find All: Searches the entire file for events similar to the current selection.
 MARKERS & REGIONS 195

63. Markers & Regions

Overview
Markers and Regions allows you to define and save particular points or selections in time for
your audio file. All markers and regions that are created will be saved with your audio file when
exporting.

Controls

• Checkboxes: The checkboxes to the left of your Marker list designate your selection. The
checkbox at the top of the list toggles between selecting all and none.
• Play Button: Begins playback from the start point of your marker or region.
• Find Button: Moves the playhead to the exact position of your marker, but will not begin
playback. For a region, Find will select the audio in the region.
• Add: Creates a new marker point at the exact current position of the playhead.
• Remove Selected: Deletes the desired marker or region.
• Select all: Selects all markers and regions.
• Select None: Deselects all markers and regions.
• Import Marker File: Opens exported marker information from a file.
• Export Marker File: Saves markers and regions in a tab-delimited text file that can be opened
and used in another file or session of RX.

More Information
• You can add a marker or region at the current location of the playhead by pressing the M key or
selecting Add Marker or Region from the Window menu.
• Markers and regions can also be created from the Markers and Regions window.
• If you have a large number of markers and regions, you may use the search box in the upper
right to quickly locate the ones you’re looking for.
 SPECTRUM ANALYZER 196

64. Spectrum Analyzer

Overview
A spectrum analyzer uses a fast Fourier transform (FFT) to extract frequency information from a
waveform. Depending on the size of the FFT, the signal energy of thousands of frequency bands
can be visually represented on a graph.
The RX Spectrum Analyzer will show the momentary spectrum of audio around the current
playhead position, the average spectrum of a selected time and frequency range, or the real-time
spectrum of the audio at the output of RX’s playback.

Controls

Channel View Selection Buttons

When viewing the spectrum analyzer of a stereo file, you can toggle the Left or Right channel
spectrum display on or off by clicking on the L or R buttons in the upper right hand corner of the
spectrum.

Peak Finding
The RX Spectrum Analyzer has a peak-finding feature that automatically detects peaks in the
spectrum data. If you hover your mouse cursor near a peak in the spectrum, a readout will
appear displaying the exact frequency of the peak, its amplitude, and the closest musical note.
This peak-finding readout can provide much higher accuracy than simply inspecting the graph by
zooming in on the display and/or increasing the FFT size in the settings window.
The circle displays the exact amplitude and frequency of the spectral peak. It is usually slightly
above the spectrum, because each spectral peak consists of several FFT bins, and their power
is added together. This effect is known as spectral smearing (or frequency smearing) and is
controlled by the choice of a weighting window.
 WAVEFORM STATISTICS 197

65. Waveform Statistics

Overview
The Waveform Statistics window shows information about the waveform within your current
selection. This information can help you find and fix problems like clipping more easily. It is also
useful for comparing two similar selections or files. The Waveform Stats window will update to
display information about your active track and can remain open as you switch between track tabs.
• If no selection is made in the active file tab, the Waveform Statistics will populate with values
based on the entire file.
• Values that appear in the Waveform Statistics window are based on the full bandwidth of the
selection. This means that the Waveform Statistics are gathered based on time selections,
and will display values for all frequencies in a time selection, even if the current selection only
includes a specific frequency range.
• Selecting the cursor icon next to a value readout will move the playhead to the position in the
file where the level was detected.

Controls

• TRUE PEAK LEVEL: The highest peak level detected, including signal levels between digital
samples (called ISPs or Intersample Peaks).
• SAMPLE PEAK LEVEL: The maximum level of digital samples in your selection.
• MAX. RMS LEVEL: The highest RMS level detected in the selection.
• MIN. RMS LEVEL: The lowest RMS level detected in the selection.
• TOTAL RMS LEVEL: The RMS level of the entire selection.
• POSSIBLY CLIPPED SAMPLES: The number of samples where the true peak signal level
exceeds 0 dBTP.
 WAVEFORM STATISTICS 198

• DC OFFSET: The amount of DC offset, in percent of the full scale. Hover this value with your
cursor to see this level in decibels.
• MAX. MOMENTARY LOUDNESS: Momentary loudness is computed on a K-weighted audio
signal using 400-ms windows, as defined by the BS.1770 specification.
• MAX. SHORT-TERM LOUDNESS: Short-term loudness is computed using 3000-ms windows,
as defined by the BS.1770 specification.
• INTEGRATED LOUDNESS: Displays the integrated loudness level (as defined by BS.1770-2).
Settings in the Loudness Control module may switch this field to BS.1770-1 standard.
• LOUDNESS RANGE (LRA): Displays the loudness range (as defined by BS.1770). This value
reflects the dynamics of audio levels in the selection.

More Information

True peak
True peak reflects the expected peak level of the analog waveform after digital-to-analog
conversion. In practice, it is calculated by oversampling of the digital waveform according to
BS.1770-3 standard. Different software may measure true peak levels slightly differently, because
the standard leaves some freedom in choice of oversampling filters.

RMS
RX measures RMS using 50-ms windows (more specifically, a Hann window with a 100-ms period)
and references levels using either an AES-17 standard (full-scale sine wave = 0 dB RMS) or a
“scientific”standard (full-scale square wave = 0 dB RMS). These RMS measurement standards
differ by 3 dB and can be chosen in RX Preferences.
 RX PLUG-INS 199

66. RX Plug-ins

Plug-in Formats
The RX 11 installer will install only the following plug-in formats:
• AU
• AAX (Realtime & Audiosuite)
• VST3
• AU ARA (Spectral Editor ARA & Music Rebalance ARA - Compatible with Logic Pro only)
All plug-in formats are 64-bit only. VST2 is no longer supported.
RX 11 Ambience Match, Dialogue Isolate, and De-rustle plug-ins are available as AAX Audiosuite
plug-ins in Pro Tools only.

Learning in Audiosuite
When using RX AudioSuite plugins in Pro Tools, there are two main workflows for learning noise
profiles. Each has its own benefits and drawbacks. These two workflows are outlined as follows:

Method 1 - AS Learn (Audiosuite Learn)

Make a selection and click the Learn button that appears in the bottom right of the AudioSuite
plugin wrapper. With this method, learning happens offline, saving you time.

It is important to note that learning with this method also includes “handles” in the learn pass.
Handles are a Pro Tools feature which extend rendered selections by a set amount. This creates a
safety net of additional processed material underneath the clip edges to allow for flexibility during
editing. By default, Pro Tools handles are set to 2.00 seconds. With this handle setting, for a
given selection, the learn pass would also include audio 2 seconds before and 2 seconds after the
selection (pictured below).
 RX PLUG-INS 200

This may account for unexpectedly intense learned noise profiles, because the plugin is receiving
noise + dialogue or other material on either side of the noise selection. To avoid this, you can set
handles to 0 via the text box next to the Render button, or learn in the middle of a long section of
noise.
If you regularly use handles in your workflow and want to avoid manually turning handles on and
off, then Method 2 below may provide better results for you.

Method 2 - UI Learn
1. Make a selection and activate the Learn button that is inside the RX plugin UI, usually on the
upper left side.
2. While the learn state is active & listening (button turns blue), click Preview (button with speaker
icon) in the lower left corner of the AudioSuite plugin wrapper. This passes the selected audio
into the learn pass in realtime.
3. Stop transport when you are finished learning, and the learn pass will turn itself off
automatically.

A benefit to this method is that only audio you’ve selected will playback into the plugin. There is
no hidden audio provided to this learning pass via Pro Tools’ handles feature. The drawback is that
you must listen to your learn pass in real time. This “UI Learn” method may provide better results if
you typically work with handles on (which is common in post-production).

Presets
From the Preset Manager, you can select from default presets and presets you have saved.
To browse presets, press the Presets button and click the name of any preset. If you like what you
hear, press the Presets button again to hide the window.
 RX PLUG-INS 201

• ADD: Clicking this button adds the current settings as a new preset. You can type a name and
optionally add comments for the preset.

Note that characters that are not allowed in Windows file names such as * or /
cannot be used as preset names and will be ignored when entered. This is to
ensure consistency with the names of the underlying .xml files for easy backup
and transferring.

• REMOVE: To permanently delete a preset, select the preset from the list and click the Remove
button.
• UPDATE: When you click the Update button, your current settings before you opened the preset
window are assigned to the selected preset (highlighted). This is useful for selecting a preset,
tweaking it, and saving your changes to the existing preset.
• IMPORT: Imports a preset into the preset folder.
• FOLDER: Opens a dialog that shows your current preset folder. You can also select a new
preset folder from this dialog.
• RENAMING PRESETS: You can double click on the name of a preset to enter the edit mode and
then type a new name for that preset.
• CANCEL: Press Escape to close the preset system dialog and revert to the settings when you
opened the preset manager.

History
Pressing the History button opens the History window. This view gives you a list of all of the
actions you’ve performed inside the plug-in, allowing you to step back to previous settings and
undo changes.
• CLEAR: Resets the History list.

I/O Meters
• Select plug-ins feature an input and output gain control and input and output metering.
• In stereo instances of the plug-ins, the gain controls are linked stereo gains.

Options

General Options

Authorization & Updates

The Authoriztion and Updates section of the General Options tab includes the following options:
• Authorization: Learn more in the Authorization chapter.

Host Performance
• Enable Multicore (Spectral De-noise only) This option is available in the Spectral De-noise
plug-in. Enabling this option lets RX process Spectral De-noise Quality Modes C and D more
efficiently by spreading its processing across multiple computer cores.
 RX PLUG-INS 202

I/O Options
The RX I/O meter displays a lower bar representing the average level (RMS) and a higher bar
representing peak level. There is also a moving line above the bar representing the most recent
peak level or peak hold.
• PEAK HOLD TIME: If peak hold is on, you can choose different peak hold times. The choices
are 250 ms, 500 ms, 1000 ms, 5000 ms and Infinite. If set to infinite, the peak value will be held
until you click on the current peak display.
• INTEGRATION TIME: Specifies the integration time for RMS calculation. In most RMS meters,
the integration time is set to around 300 ms, which makes the RMS meter similar in ballistics to
VU meters.
• READOUT: Allows you to control what level value is displayed on top of the meters: peak or
actual (real time). If set to Max Peak, the display will show the highest peak level. If set to
Current, the display will reflect the meter’s current value.
• ENABLE I/O METERS: Turns the level meters on or off.
• SHOW PEAK HOLD: Turns the peak hold display for the level meters on or off.

Latency
Some of the processing modes in the RX plug-ins are very CPU intensive and result in a delay of
the signal. That is, RX needs some time to process the audio before it can send it back to the host
application. That time represents a delay when listening or mixing down.
Most modern DAWs and NLEs provide delay compensation—a means for RX plug-ins to tell the
application it has delayed the signal, and the host application should “undo” the delay on the track
(usually by adding compensating delays to other tracks in real-time processing, or by adjusting the
rendered file after processing in offline processing). When Enable Delay Compensation is enabled
in the Latency menu, we will report our latency to the host application.
If your application doesn’t support it, or skips/stutters with this option on, you can always manually
correct the delay offset in the host application (manually edit out the short delay of silence). To
help you perform manual correction, the delay RX introduces is shown below as Total System
Delay in both samples and milliseconds.
 RX MONITOR 203

67. RX Monitor
Plug-in only

Overview
When using the RX Connect plug-in, some DAW/NLEs monopolize the system’s audio drivers,
preventing RX from playing audio through the same output device. The RX Monitor plug-in allows
you to listen to the output of the RX Audio Editor through the audio driver output of your host
application. This is particularly useful when using RX Connect with a host application that would
need to be closed for the RX Audio Editor to access the output driver it is using.

Workflow
1. Insert the RX Monitor plug-in on an Aux or Instrument track in your DAW/NLE. When you
first insert the RX Monitor plug-in, you will see a Status: Disconnected message. You need
to configure RX Monitor as the Audio Driver in the RX Audio Editor to change the status to
Connected.

2. Change the Audio Driver in the RX Audio Editor:

a. Open Preferences.
b. Select the Audio tab.
c. In the Driver Type dropdown menu, select RX Monitor.

3. Navigate back to your DAW and confirm the RX Monitor plug-in is showing a “Connected”
status.

4. Playback audio in the RX Audio Editor, the output will be routed to the track you added RX
Monitor to in your DAW/NLE.

NOTE
If you are running virus protection software or a firewall on your DAW, you may need to
grant permission for RX Monitor to run.
 RX CONNECT 204

68. RX Connect
Plug-in Only

Overview
The RX Connect plug-in sends a clip, or multiple clips, to the RX 11 standalone application for
editing and repair. This gives you access to all of RX 11’s modules in one place, and provides
the benefits of RX’s offline processing and visual interface. RX Connect is available from the
AudioSuite menu in Pro Tools, or as an AU or VST3 plug-in from your DAW’s effects menu.

Controls

There are two modes for using RX Connect:

• Send For Reference: This is meant for analysis only. The clips are imported into RX 11 but
cannot be sent back to your host.

NOTE ABOUT REFERENCE MODE

This mode of RX Connect will not open RX 11 Audio Editor automatically. Opening
the Audio Editor after using send for reference will reveal the file in the RX Audio
Editor.

• Send For Repair: Selected clips are sent to the RX 11 Audio Editor for repair, and you can send
them back to your host from the RX Audio Editor.

MORE INFORMATION
For more information on using RX Connect in different DAWs or NLEs, please
refer to the following sections, or check out the RX Connect FAQ on our Support
website.

DAW/NLE specific instructions

The following sections outline host specific instructions for using RX Connect to Send audio to the
RX Audio Editor and back to your host application:
 RX CONNECT 205

Adobe Audition CC RX Connect Workflow

1. Inside of Audition, select the Waveform view.
2. Highlight the area of audio that requires editing.
3. In the Effects menu, load the RX 11 Connect plug-in from VST3 > Restoration > iZotope (If you
do not see the RX 11 Connect plug-in, open the Audio Plug-in Manager and Scan for Plug-ins,
then make sure RX 11 Connect is enabled).
4. When the plug-in window opens, click Apply.
5. RX 11 will automatically load. Perform your desired audio edit, then click SEND BACK to send
the audio back to Adobe Audition. The Waiting for Connect message will appear.
6. Re-load the RX 11 Connect plug-in from the Effects menu. It will now display a message Press
Apply to commit changes.
7. Click Apply to apply the audio edit from RX to your audio file in Adobe Audition.

Avid Pro Tools RX Connect Workflow

1. Choose the audio to be sent to the RX Audio Editor by selecting the audio clip(s) in the timeline
that you want to edit, and opening RX Connect from the AudioSuite Noise Reduction menu.
2. If you just need to load a noise profile or analyze some audio, choose Reference to send the
audio one-way, but for the complete round-trip workflow click Repair and then hit Send. You’ll
see this opens the audio in the RX Audio Editor.
3. With HDX systems, Pro Tools will have control of your audio drivers, so you aren’t able to
hear the output of the RX Audio Editor. However, the RX Monitor tool is built to solve just this
problem. In Pro Tools, create a dedicated aux track for monitoring RX, and insert RX 11 Monitor
from the Noise Reduction or Sound Field menus,
4. Then, go to the ‘Preferences’ menu in the RX Audio Editor by clicking on the wrench icon in the
top-right of the window. In the Audio tab, set your Driver type to be RX Monitor. Now we can
hear the output of the RX Audio Editor through your Pro Tools output chain.
5. After you’ve made the desired edits in RX, click Send back at the top of the window. Click
Render in the RX Connect window, and the repaired audio will be placed back into your session.

Some engineers might choose to create duplicate playlists before making any repairs
to their audio, but you can ‘undo’ these RX Connect changes just like any AudioSuite
process. If you make extensive repairs inside of the RX Audio Editor, you can also save
an .rxdoc of the file, which will preserve all your adjustments so you can modify them
later if you need to.

Audiosuite modes

When using Audiosuite plug-ins, there are various user definable input and output options, which
affect how you may use RX Connect. These options are:
 RX CONNECT 206

Input
• Clip-by-clip: Recognizes individual clips in the timeline, as well as fades.
• Entire selection: Treats the entire selected area as one clip.

Input channel modes

• Mono mode: Treats mono, dual mono and stereo clips, as well as multi-channel clips, all as
discrete mono clips (e.g. a stereo clip will send as two separate mono files).

MONO MODE NOTE

Please note, this can result in large groups of audio clips being sent to RX,
potentially exceeding the maximum file limit of 32.

• Multi-input mode Treats dual mono and stereo audio clips as one entity.

Output
• Overwrite files: Destructive processing of the audio clip(s) in the session, overwriting the
original file with the new file sent from RX.
• Create individual files: Nondestructive processing of the audio file(s) in the session, replacing
them with the audio processed in RX. This mode preserves individual clips and fades/handles.
• Create continuous files: Nondestructive processing of the original audio file. Creates a new
audio file with the audio sent back from RX, consolidated into one continuous clip.

Steinberg Cubase and Nuendo RX Connect Workflow

1. Select the audio clip you wish to apply changes to.
2. Navigate to the Audio menu in Cubase/Nuendo and select Direct Offline Processing.
3. Uncheck Auto-Apply in the Direct Offline Processing Window.

1. Click the + Plug-in button and select RX 11 Connect.

2. Press Apply to send the file to RX.
3. The RX Audio Editor application will automatically open with the file you sent loaded in a tab
named “Cubase 1” or “Nuendo 1”.
4. Make the desired changes to your file in the RX Audio Editor.
5. Click the Send back button in the RX file tab display to send the updated file back to Cubase/
Nuendo.
 RX CONNECT 207

1. Click the Apply button in the Direct Offline Processing window to apply the changes to the file
in the session.

APPLY BUTTON MISSING?

If Auto Apply is enabled in the Direct Offline Processing window, the Apply button
will not be available. Disable Auto Apply and restart the process with a new
instance of RX Connect.

2. Navigate to the Audio menu and select Make Direct Offline Processing Permanent.

Using RX as an External Audio Editor

Some hosts don’t support the use of RX Connect for round-trip editing, please refer to the
instructions below for host specific workflows.

Adobe Premiere Pro CC with RX as an external audio editor

1. Inside of Premiere, right-click on an audio clip in your timeline and select Reveal in Finder (Mac)
or Reveal in Explorer (Windows).
2. Open the resulting file in the RX 11 Audio Editor.
3. Perform desired processing in the RX application.
4. When you have made the desired changes to your file, go to the RX File menu and select
Overwrite Original File.

NOTE
If you have Adobe Audition installed as well, you can right-click on an audio clip in your
timeline, and select Edit Clip In Adobe Audition. Then follow these steps for using RX
Connect with Adobe Audition.

Apple Logic Pro X with RX as an external audio editor

RX is a powerful audio editor that Apple Logic Pro X users can use to get better sounding audio. To
use RX with Logic, you must first set it up as an external audio editor.

How to set up RX as an external audio editor

1. Open Logic Preferences > Advanced.
2. Under Additional options, enable the Audio check box.
3. In Preferences, click the Audio tab and select the Audio File Editor tab.
4. Under Audio File Editor, click on the External Sample Editor to select iZotope RX 11 Audio
Editor from your applications folder.
 RX CONNECT 208

Workflow
1. Select the clip you wish to edit in your timeline.
2. Select Edit > Open in iZotope RX 11 Audio Editor, Shift + W.
3. The file will open in RX 11. Once you’ve completed your edits, open the RX 11 File menu and
select Overwrite Original File.
4. Close the tab, navigate back to Logic Pro X and wait for the waveform to update.

DaVinci Resolve with RX as an External Editor

To use RX as an external audio editor with Resolve, you must first set it up as an external process
in the Resolve Preferences menu.

Configure RX as an external process in DaVinci Resolve

1. Open the DaVinci Resolve menu and select Preferences
2. Go to the Audio Plugins tab
3. Click Add in the Setup External Processes section
4. Configure the New External Process:
a. Name: Double-click on the default name of the new external process to edit it. Edit the text
and hit return to save the changes. We recommend changing this name to ‘RX’ so it is easily
identifiable in context menus
b. Path: Double-click on the Path field to set the path to the RX Audio Editor application. In the
system dialog that appears, navigate to your Applications/Program Files directory, select
the iZotope RX Audio Editor application, and click “Open”
c. Type: Use the default setting of “Command Line”
5. Click Save in the bottom of the Preferences window and click OK in the ‘Preferences Updated’
dialog that appears

Working with RX as an external editor with DaVinci Resolve

Once you have configured the RX application as an external process in Resolve, use the following
steps to send clips to RX for editing:
1. Open the Fairlight tab (musical notes icon at the bottom of Resolve)
2. Right-click on the audio clip you want to edit in RX
3. Go to External Audio Processes in the context menu, select the RX process you configured in
Resolve Preferences
4. The clip will open in the RX application, make your edits, and save your changes in RX (using
File menu > Save, or Command+S /ctrl+S) to update the clip in your Resolve project

NOTE
When you send a clip from Resolve to RX, saving changes in RX will immediately
update the clip in your Resolve project. You may want to create a copy of the clip
in your Resolve project before opening and editing it in RX if you want to maintain
access to the original.
 RX SPECTRAL EDITOR 209

69. RX Spectral Editor

Overview
The Spectral Editor plug-in combines ARA 2.0 technology with the award-winning DSP and
selection-based editing workflows of the RX Audio Editor. It allows you to identify and reduce
undesirable sounds in your tracks without leaving your DAW.

Getting Started with Spectral Editor (ARA)

Logic Pro
Spectral Editor is currently only available in Logic Pro as an ARA 2.0 plug-in (Rosetta only). You can
add RX Spectral Editor (ARA) to your project using the following steps:
1. Make sure you are using the latest version of Logic Pro. The RX Spectral Editor ARA plug-in is
supported in Logic Pro 10.7+ or higher. We recommend that you update Logic Pro to the latest
available version to ensure the best experience using RX Spectral Editor (ARA).
2. Add RX Spectral Editor (ARA) as the first insert on an Audio track.

3. Select a clip in Logic and start playback. The selected audio will appear in the spectrogram view
of the Spectral Editor (ARA) plug-in after starting playback.
4. Make a selection in the spectrogram view of RX Spectral Editor.

5. Process your selection using one of the following modes described below.
6. To work on another clip in your session, select it and start playback. Once playback has started,
you will see your new clip load in the spectrogram. Trimmed clips appear the same way in the
spectrogram as they do in Logic.
Pro Tools
 RX SPECTRAL EDITOR 210

You can add RX Spectral Editor (ARA) to your Pro Tools project using the following steps:
1. Make sure you are using the latest version of Pro Tools. The RX Spectral Editor ARA plug-in is
supported in Pro Tools 2024.6 or higher.
2. Clip menu:
a. Select a clip on the timeline
b. Open the Clip menu at the top of Pro Tools
c. Select RX Spectral Editor > Edit to open that clip in the Spectral Editor.
3. You can add Spectral Editor to an entire track in one of two ways:
a. The Elastic Audio/ARA plug-in selector drop down on the track itself
b. Track menu
i. Select a track
ii. Go to Track Menu at the top of Pro Tools
iii. Select RX Spectral Editor > Edit.
iv. Once added to a track, select a clip on the track to open it in Spectral Editor.
All clips that exist on a track at the time Spectral Editor is added to it will be visible to the editor. If
you add additional clips to the track or process pre-existing clips with AudioSuite, you will need to
manually enable Spectral Editor on those new clips with the Clip menu method above.
Closing & Reopening the Editor
You can close and reopen the Spectral Editor window by clicking on the insert that appears in the
Elastic / ARA slot on a track. You can also access it via the RX Spectral Editor tab at the bottom of
the Pro Tools Edit window.
Committing edits to a new file on disk
When you’re finished making edits with Spectral Editor, if you prefer, you can commit your edits
directly to a new audio file on disk. This is similar to an AudioSuite workflow. To accomplish this,
right click on a clip you want to commit and select RX Spectral Editor > Render. From this same
menu, you can also clear or bypass Spectral Editor ARA processing.
If you edit a clip with Spectral Editor ARA and then process that clip with an AudioSuite plugin, your
edits will be committed to the new file.
Studio One
The VST3 version of Spectral Editor ARA works directly on clips as an “Event FX” in Studio One.
Event FX are applied to individual audio events rather than as inserts on entire tracks.

MACOS NOTE
Although an AU version of Spectral Editor ARA will show up as available in Studio One,
it is not supported. The UI will direct you to use the VST3 version.

You can add RX Spectral Editor (ARA) to an audio event in your Studio One session using one of
the following methods:
1. Make sure you are using the latest version of Studio One. The RX Spectral Editor ARA plug-in is
supported in Studio One 6.0 or higher. Studio One 6.5 introduced additional improvements that
make it easier to access ARA plugins.
 RX SPECTRAL EDITOR 211

2. Event FX Context menu [Studio One 6.5 or higher]:

a. Right click on an audio event
b. In the context menu, select Event FX > RX 11 Spectral Editor
3. Effects Browser:
a. [Studio One 6.5 or higher], Spectral Editor will appear in the “Event Editors” category at the
top of the Effects browser.
▶ Click & drag RX 11 Spectral Editor onto an audio event
b. In Studio One 6 versions before 6.5, Spectral Editor will appear under the regular VST3
location.
i. While dragging Spectral Editor onto your event, if the status text doesn’t say “Add Event
FX”, then hold OPT (Mac) or ALT (Win) and that message should appear.
ii. The plugin will display a warning message if it has been incorrectly added as a normal
channel effect.
4. Inspector:
a. Open the Inspector (the ‘i’ icon in the top left)
b. Select an audio event
c. Underneath the Mixer section of the inspector, there is a section with info about the
currently selected event.
d. Click the Event FX Enable button, which will open a toggle dropdown
e. In the Inserts section that appears, use the + sign to add Spectral Editor directly to the clip
as an Event FX.
Reopening the Editor
To re-open the Spectral Editor window on a clip after you’ve closed the UI, either
• Select the clip and double click the Spectral Editor Event FX insert at the bottom of the
Inspector
• [Studio One 6.5 or higher] Right click on the clip and select Event FX > RX 11 Spectral Editor
from the context menu again.
Committing edits to a new file on disk
When you’re finished making edits on a clip with Spectral Editor, if you prefer, you can commit your
edits directly to a new audio file on disk. You do this with the Render command. With the Inspector
open, (‘i’ icon in the top left), clicking on a clip will display info about that clip at the bottom. Click
the orange Render button under the Event FX section at the bottom. You can also click Restore if
you wish to bring the Editor back after Render.
A note about duplicated/copied audio
In Studio One, all copies of the same source audio will inherit the same edits when Spectral Editor
is applied to each copy. For example, if there are two copies of Clip A in your session and you add
the Spectral Editor plugin to each, then any processing applied to copy 1 will be synchronized to
copy 2, and vice versa.
If you wish to decouple edits for multiple copies of the same source audio, right click on your clip
and select Audio > Audio Parts > New Clip Version. You will see a small number appear in the
bottom left of your clip. Clips with the same version number will share edits, as long as Spectral
Editor ARA is applied on all the clips. In the session Pool, all clip versions will be grouped under the
parent clip and can be swapped in/out from there.
Consult Studio One’s help documentation for more information about Clip Versions and working
with Event FX.
 RX SPECTRAL EDITOR 212

PLAYBACK IS REQUIRED TO REFLECT CHANGES IN SPECTROGRAM

ARA plug-ins are updated when playback is started. If you trim or otherwise change
a clip that was already loaded in Spectral Editor, the spectrogram will not be updated
until you’ve stopped and restarted playback with the clip selected.

Troubleshooting
If you find that RX Spectral Editor (ARA) is missing from the plug-in selection menu, try the
following:
1. Ensure you are adding the RX Spectral Editor (ARA) plug-in to the first insert slot of an Audio
track.
2. Clear the AudioUnit Cache and rescan plug-ins:
• Logic 10.7 or higher: Go to the Logic Pro menu > Preferences… > Plug-in Manager… and
select “Full Audio Unit Reset” at the bottom of the window to reset the cache and rescan
plug-ins.
3. See the ARA Requirements and Limitations section below for more information about working
with Spectral Editor (ARA) in Logic Pro.

Controls
Once your clip is loaded in the spectrogram view of the Spectral Editor, you’ll have access to the
following controls:

Processing Mode
Select Attenuate, Replace, Gain, De-click or De-hum and the relevant processing parameters will
appear on the right side of the plugin.

Selection Tools
Determines the selection tool type used when making audio selections in the spectrogram.

See the Interactive Tools chapter for more information about selection tools and modifiers.

General Controls
 RX SPECTRAL EDITOR 213

• Undo: Reverts to the previous processing step.

EDITING COPIES OF CLIPS

The Undo function in Spectral Editor will apply to all non-unique copies of a clip.
If you wish to edit different copies of the same clip independently using Spectral
editor, please bounce them first.

UNDO HISTORY WARNING

Undo history is cleared when a session is saved.

• Redo: Re-applies the next processing step in your history.

• Reset: Reverts all changes in the audio to the unprocessed state.
• Settings: Opens the plug-in General Settings menu.
• Help: Opens the installed HTML help documentation in your default web browser.

USE THE TRACKPAD TO NAVIGATE THE SPECTROGRAM

Scroll with the trackpad vertically to zoom in and out, and horizontally to scroll
forward/backward in time.

Processing Modes

Attenuate

Attenuate removes sounds by comparing the content inside of your selection to the content
outside of your selection. It modifies dissimilar audio in your selection to be more similar to the
surrounding audio. Attenuate does not re-synthesize any audio.
 RX SPECTRAL EDITOR 214

• Strength: Adjusts the amount of attenuation.

• Region: Defines how much of the surrounding content will be used for interpolation.
• Weight: Gives more weight to the surrounding audio before or after the selection.
• Direction: Determines where the material used in the repair process is located in relation to the
current selection. Note: Replace mode always uses Horizontal interpolation when processing a
selection.
• Horizontal: Signal to the left and right of the current selection will be used for interpolation.
• 2D: Signal above, below, to the left and to the right of the current selection will be used for
interpolation.
• Vertical: Signal above and below the current selection will be used for interpolation.
• Process: Applies processing to the current selection.
• Resize: Allows you to click and drag to change the size of the plug-in window.

Replace

Replace resynthesizes badly damaged sections (such as gaps) in tonal audio. It completely
replaces the selected content with audio interpolated from the surrounding data.
 RX SPECTRAL EDITOR 215

• Strength: This control is not active in Replace.

• Region: Defines how much of the surrounding content will be used for interpolation.
• Weight: Gives more weight to the surrounding audio before or after the selection.
• Direction: This control is not active in Replace. The Replace mode always uses Horizontal
interpolation when processing a selection.

PROCESSING SELECTIONS LONGER THAN 10 SECONDS PROCESSING SELECTIONS

LONGER THAN 10 SECONDS
Attenuate: Horizontal, Attenuate: 2D, and Replace mode do not support processing
selections longer than 10 seconds. If your selection is longer than 10 seconds, the
processing will automatically default to Attenuate: Vertical.

Gain

Gain boosts or cuts the level of the signal by the designated decibel amount.

De-click

De-click analyzes audio for amplitude irregularities and smoothes them out. Use De-click to
remove a variety of short impulse noises, such as clicks caused by digital errors, mouth noises,
and interference from cell phones.
 RX SPECTRAL EDITOR 216

• SENSITIVITY: Determines how many clicks are detected in the signal.

• ALGORITHM: Provides options for adjusting the configuration and processing quality of click
interpolation, depending on the types of clicks and pops present in the audio.
• SINGLE BAND: Processes quickly and works well on very narrow “digital” clicks
• MULTI-BAND (PERIODIC CLICKS): Uses multi-band processing for removing regularly
repeating clicks with a wider spectrum, or regular clicks that have concentrated low or
high energy (like thumps or optical soundtrack perforation noise)
• MULTI-BAND (RANDOM CLICKS): Uses multi-band processing for wider vinyl clicks and
thumps, with a protective algorithm for preserving periodic audio elements characteristic to
certain instruments such as brass or vocals
• LOW LATENCY: Works well on mouth clicks and other clicks that cannot be handled by
other algorithms. Low latency mode is suitable for real-time work.

De-hum

De-hum is designed to remove persistent tonal noise, like AC noise, electrical buzz or interference
introduced by EMF (Electro-magnetic fields). De-hum uses the Dynamic mode with Adaptive
learning enabled.

• SENSITIVITY: Adjusts the amount of hum that will be removed.

• BANDS: Controls the number of dynamic notch filters, which update their gains based on the
input signal to avoid ringing. Adjust to higher values for complex hums, at the expense of
transparency.
 RX SPECTRAL EDITOR 217

• FILTER Q: Controls the bandwidth of the notch filters.

Working with Comp Takes in Logic Pro

Spectral Editor supports comp takes in Logic. 7.5+ (Rosetta only). To process comp takes, the
individual takes that make up the comp must be selected and processed.

The Logic comp region at the top will not appear in the spectrogram if selected. Choose the
individual takes and toggle playback to work on comps in Spectral Editor.

Keyboard Shortcuts
The following keyboard shortcuts are available in Spectral Editor. CMD is for Mac, CTRL is for
Windows:
• Undo in Spectral Editor: CMD/CTRL+Z
• Redo in Spectral Editor: Shift+CMD/CTRL+Z
• Switch to Attenuate mode: CMD/CTRL+1
• Switch to Replace mode: CMD/CTRL+2
• Switch to Gain mode: CMD/CTRL+3
• Switch to De-click mode: CMD/CTRL+4
• Switch to De-hum mode: CMD/CTRL+5
• Process: CMD/CTRL+Enter
• Choose Time selection tool: 1
• Choose Time-frequency selection tool: 2
• Choose Frequency selection tool: 3
• Choose Lasso selection tool: 4
• Choose Brush selection tool: 5
• Choose Magic wand selection tool: 6

ARA Requirements and Limitations

• Spectral Editor (ARA) must be the first plug-in on a track.
 RX SPECTRAL EDITOR 218

• Spectral Editor (ARA) cannot be added to a software instrument track or bus.

• Spectral Editor (ARA) is not supported by Selection Based Processing.
• Host presets are not supported by Spectral Editor (ARA).
• Automation is not supported by Spectral Editor (ARA).
• Playback must be started to gather audio data. After editing, playback must be stopped and
started again to update data.
• Spectral Editor (ARA) cannot process Apple Loops, compressed audio, flex audio, or reversed
regions. These regions can be bounced in place in order to be processed by Spectral Editor
(ARA).
• Spectral Editor (ARA) will output silence for stereo files on mono tracks.
• Spectral Editor (ARA) will output silence for regions whose sample rate does not match the
sample rate of the session.
 KEYBOARD SHORTCUTS 219

70. Keyboard Shortcuts

The RX Audio Editor includes options for defining custom keyboard shortcut commands.
The column named “RX Shortcut Command Name” in the table below lists the underlying name
associated with each default shortcut in the RX Audio Editor. These are useful to know when
assigning custom shortcuts in the Keyboard tab.

IMPORTING KEYBOARD SHORTCUTS FROM RX 10

If you would like to carry over custom shortcuts you set in RX 10, you will need to
export the keybindings from RX 8 and then import that keybindings file to RX 11 using
the Import option in the Keyboard tab.

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
New… command+N ctrl+N File.New
New from shift+command+N ctrl+shift+N File.NewFromClipboard
Clipboard
Open… command+O ctrl+O File.Open
Save command+S ctrl+S File.Save
Save As… shift+command+S ctrl+shift+S File.SaveAs
Save RX File.SaveRXDocument
Document
Save RX File.SaveRXDocumentAs
Document
As…
Overwrite option+command+S ctrl+alt+S File.SaveOverwriteOriginal
Original File
Export… command+E ctrl+E File.Export
Export shift+command+E ctrl+shift+E File.ExportSelection
Selection…
Export option+command+E ctrl+alt+E File.ExportRegions
Regions to
Files…
Close command+W ctrl+W File.Close
Close All shift+V ctrl+shift+W File.CloseAll
Zoom out command+0 ctrl+0 Zoom.AllOutFull
full all rulers
Zoom in on shift+up arrow shift+up arrow Zoom.AmpIn
amplitude
ruler
Zoom out on shift+down arrow shift+down arrow Zoom.AmpOut
amplitude
ruler
 KEYBOARD SHORTCUTS 220

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Zoom in on shift+command+up arrow shift+ctrl+up arrow Zoom.FreqIn
frequency
ruler
Zoom out on shift+command+down shift+ctrl+down Zoom.FreqOut
frequency arrow arrow
ruler
Zoom in on up arrow up arrow Zoom.TimeIn
time ruler
Zoom in on command+= ctrl+= Zoom.TimeIn
time ruler
Zoom on left command+[ ctrl+[ Zoom.TimeLeftEdge
side of time
ruler
Zoom out on command+- ctrl+- Zoom.TimeOut
time ruler
Zoom out on down arrow down arrow Zoom.TimeOut
time ruler
Zoom out shift+command+- ctrl+shift+- Zoom.TimeOutFull
full on time
ruler
zoom on command+] ctrl+] Zoom.TimeRightEdge
right side of
time ruler
Zoom to command+\ ctrl+[ Zoom.TimeSelection
time
selection
Undo command+Z or ctrl+Z or ctrl+alt+Z Edit.Undo
option+command+Z
Redo command+Y or ctrl+Y or ctrl+shift+Z Edit.Redo
shift+command+Z
Cut command+X ctrl+X Edit.Copy
Copy command+C ctrl+C Edit.Cut
Paste command+V ctrl+V Edit.Paste
Paste option+command+V ctrl+alt+V Edit.PasteInsert
Special>
Insert
Paste option+shift+command+V ctrl+alt+shift+V Edit.PasteReplace
Special>
Replace
Paste shift+V shift+V Edit.PasteMix
Special> Mix
Paste option+V alt+V Edit.PasteMixInvert
Special>
Invert and
Mix
 KEYBOARD SHORTCUTS 221

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Paste option+shift+V alt+shift+V Edit.PasteToSelection
Special> To
Selection
Only
Paste shift+command+V ctrl+shift+V Edit.PasteClipGainOnly
Special> Clip
Gain Only
Deselect command+D ctrl+D Edit.Deselect
Reselect shift+command+D ctrl+shift+D Edit.Reselect
Select All command+A ctrl+A Edit.SelectAll
Invert shift+command+I ctrl+shift+I Edit.SelectInverse
Selection
Invert command+I ctrl+I Edit.SelectInverseFreq
Selection
Frequencies
Select shift+command+H ctrl+shift+H Edit.SelectHarmonicsByNumbers
Harmonics…
Begin [ [ Edit.SetSelectionStart
Selection At
Playhead
End ] ] Edit.SetSelectionEnd
Selection At
Playhead
Delete Del Del Edit.SilenceDelete
Selection
Trim to command+T ctrl+T Edit.TrimToSelection
Selection
Snap shift+command+; ctrl+shift+; View.ToggleSnapping
Find Similar command+F ctrl+F Edit.FindSimilarEvent
Event
Window
Find Next shift+command+F ctrl+shift+F Edit.FindNextSimilarEvent
Similar
Event
Find option+command+F ctrl+alt+F Edit.FindPrevSimilarEvent
Previous
Similar
Event
Add Marker M M Edit.AddMarkerOrRegion
or Region
Edit Cursor T T Edit.EditorCursorMode.SelectTime
Mode >
Select Time
 KEYBOARD SHORTCUTS 222

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Edit Cursor R R Edit.EditorCursorMode.SelectTimeFreq
Mode >
Select Time/
Freq
Edit Cursor F F Edit.EditorCursorMode.SelectFreq
Mode >
Select Freq
Edit Cursor L L Edit.EditorCursorMode.SelectLasso
Mode >
Lasso
Edit Cursor B B Edit.EditorCursorMode.SelectBrush
Mode >
Selection
Brush
Edit Cursor W W Edit.EditorCursorMode.SelectWand
Mode >
Selection
Wand
Edit Cursor Z Z Edit.EditorCursorMode.ZoomTime
Mode >
Zoom Time
Edit Cursor shift+Z shift+Z Edit.EditorCursorMode.ZoomTimeFreq
Mode >
Zoom Time/
Freq
Edit Cursor option+Z alt+Z Edit.EditorCursorMode.ZoomFreq
Mode >
Zoom Freq
Edit Cursor G G Edit.EditorCursorMode.GrabTime
Mode > Grab
Time
Edit Cursor shift+G shift+G Edit.EditorCursorMode.GrabTimeFreq
Mode > Grab
Time/Freq
Edit Cursor option+G alt+G Edit.EditorCursorMode.GrabFreq
Mode > Grab
Freq
Open Batch command+B ctrl+B File.BatchProcessing
Processing
window
Send Command+Return ctrl+return File.SendConnectClipsBackToHost
Connect
Clips back to
host
Discard command+delete ctrl+backspace File.DiscardConnectClips
Connect
Clips
 KEYBOARD SHORTCUTS 223

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Remove Clip shift+delete shift+backspace Editor.RemoveClipGain
Gain from
selection
Remove All shift+command+delete ctrl+shift+backspace Editor.RemoveAllClipGain
Clip Gain
Toggle command+P ctrl+P Transport.TogglePlayheadFollow
Follow
Playhead
Toggle shift+command+P ctrl+shift+P Transport.CyclePlayHeadFollowMode
Follow
Playhead
Mode >
Page /
Continuous
Show Clip command+G ctrl+G View.ToggleGainCurveOverlay
Gain
Show shift+command+C ctrl+shift+C View.ToggleCompositeAudioDisplay
Channels
Separately
Show shift+command+, ctrl+shift+, View.ToggleSpectrogramSettingsVisible
Spectrogram
Settings
Decrease Shift+ Shift+ Spectrogram.FFTSizeDecrement
Spectrogram
FFT Size
Increase Shift+. Shift+. Spectrogram.FFTSizeIncrement
Spectrogram
FFT Size
Show command+ ctrl+ Edit.Preferences
Preferences
Window
Show File shift+option+command+I shift+alt+ctrl+I File.Info
Info window
Enter Full ^+command+F ctrl+^+F View.ToggleFullScreen
Screen
Exit Full esc Esc View.ExitFullScreen
Screen
Toggle I I Edit.EditorCursorMode.ToggleInstant
Instant
Process
Toggle shift+B shift+B TogglePreviewBypass
Preview
Bypass
Toggle shift+command+O ctrl+shift+O View.ToggleFloatingWindowOpacity
Window
Opacity
 KEYBOARD SHORTCUTS 224

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Toggle Input option+I alt+I ToggleInputMonitoring
Monitoring
Start or Stop Spacebar Spacebar Transport.PlayOrStop
Playback
Start or Stop shift+Spacebar shift+Spacebar Transport.PreviewOrStop
Preview
Playback
Rewind return Home Transport.Rewind
Transport
Seek to End End Transport.SeekToEnd
of file
Toggle command+L ctrl+L Transport.ToggleLooping
Looping
Toggle command+P ctrl+P Transport.TogglePlayHeadFollow
Playhead
Follow
Toggle command+R ctrl+R Transport.TogglePlayHeadReturn
Playhead
Return
Select Both shift+command+B ctrl+shift+B Editor.ChannelSelectBoth
Channels
Select Left shift+command+L ctrl+shift+L Editor.ChannelSelectLeft
Channel
Select Right shift+command+R ctrl+shift+R Editor.ChannelSelectRight
Channel
Extend shift+left Arrow shift+left arrow Editor.ExtendSelectionLeft
selection left
Extend shift+up arrow shift+page up Editor.ExtendSelectionPageLeft
selection left
by page
Extend shift+down Arrow shift+page down Editor.ExtendSelectionPageRight
selection to
the right
boundary of
current view
Extend shift+right arrow shift+right arrow Editor.ExtendSelectionRight
selection by
increment to
the right
Move option+right arrow alt+right arrow Editor.GoToNextMarkerOrSelectionBoundary
playhead to
next marker
or selection
boundary
 KEYBOARD SHORTCUTS 225

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Move option+left arrow alt+left arrow Editor.GoToPreviousMarkerOrSelectionBoun
playhead to dary
previous
marker or
selection
boundary
Nudge left arrow left arrow Editor.NudgeLeft
playhead to
the left
Nudge right arrow right arrow Editor.NudgeRight
playhead to
the right
Page Left page up page up Editor.PageLeft
Page Right page down page down Editor.PageRight
Select to shift+end shift+end Editor.SelectToEnd
End
Select to shift+home shift+home Editor.SelectToStart
Start
Process shift+R shift+R Process.Reverse
Reverse
Process shift+S shift+S Process.Silence
Silence
Process option+command+6 ctrl+alt+6 Apply.Gain
Gain
Process option+command+0 ctrl+alt+0 Apply.Leveler
Leveler
Process option+command+4 ctrl+alt+4 Apply.Loudness
Loudness
Process option+command+2 ctrl+alt+2 Apply.MatchAmbience
Ambience
Match
Process option+command+8 ctrl+alt+8 Apply.ChannelMix
Mixing
Process De- command+5 ctrl+5 Apply.DePlosive
plosive
Process De- command+2 ctrl+2 Apply.Declick
click
Process De- command+1 ctrl+1 Apply.Declip
clip
Process command+7 ctrl+7 Apply.Deconstruct
Deconstruct
Process De- command+8 ctrl+8 Apply.Dereverb
reverb
Process EQ option+command+7 ctrl+alt+7 Apply.EQ
 KEYBOARD SHORTCUTS 226

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Process EQ option+command+1 ctrl+alt+1 Apply.EQMatch
Match
Process option+command+5 ctrl+alt+5 Apply.Plug-in
Plug-in
Process De- command+3 ctrl+3 Apply.RemoveHum
Hum
Process option+command+9 ctrl+alt+9 Apply.Resampler
Resampler
Process command+6 ctrl+6 Apply.SpectralRepair
Spectral
Repair
Process option+command+3 ctrl+alt+3 Apply.TimeStretchPitchShift
Pitch
Contour
Process command+4 ctrl+4 Apply.VoiceDenoise
Voice De-
noise
Open Gain shift+option+6 shift+option+6 View.Module.ToggleGain
module
Open Leveler shift+option+0 shift+option+0 View.Module.ToggleLeveler
module
Open shift+option+4 shift+option+4 View.Module.ToggleLoudness
Loudness
module
Open shift+option+2 shift+option+2 View.Module.ToggleMatchAmbience
Ambience
Match
module
Open Mixing shift+option+8 shift+option+8 View.Module.ToggleChannelMix
module
Open De- shift+5 shift+5 View.Module.DePlosive
plosive
module
Open De- shift+2 shift+2 View.Module.ToggleDeclick
click module
Open De-clip shift+1 shift+1 View.Module.ToggleDeclip
module
Open shift+7 shift+7 View.Module.ToggleDeconstruct
Deconstruct
module
Open De- shift+8 shift+8 View.Module.ToggleDereverb
reverb
module
Open EQ shift+option+7 shift+option+7 View.Module.ToggleEQ
module
 KEYBOARD SHORTCUTS 227

Shortcut Default Mac Shortcut Default Windows RX Shortcut Command Name

Shortcut
Open EQ shift+option+1 shift+option+1 View.Module.ToggleEQMatch
Match
module
Open Plug-in shift+option+5 shift+option+5 View.Module.TogglePlugIn
window
Open De- shift+3 shift+3 View.Module.ToggleRemoveHum
hum module
Open shift+option+9 shift+option+9 View.Module.ToggleSRC
Resample
module
Open shift+6 shift+6 View.Module.ToggleSpectralRepair
Spectral
Repair
module
Open Time & shift+option+3 shift+option+3 View.Module.ToggleTimeStretchPitchShift
Pitch
module
Open Voice shift+4 shift+4 View.Module.ToggleVoiceDenoise
De-noise
module
Open option+m option+m View.ToggleMarkerPanelVisible
Markers
window
Open c c View.ToggleModuleChainVisible
Module
Chain
window
Open option+r option+r View.ToggleSpectrumAnalyzerVisible
Spectrum
Analyzer
window
Open option+d option+d View.ToggleWaveformStatsVisible
Waveform
Stats
window
Learn shift+option+command+9 shift+alt+ctrl+9 Apply.DereverbTrain
Ambience
Match
Suggest De- shift+option+command+4 shift+alt+ctrl+4 Apply.RemoveHumTrain
hum
Learn Voice shift+option+command+5 shift+alt+ctrl+5 Apply.VoiceDenoiseTrain
De-noise
Learn EQ shift+option+command+2 shift+alt+ctrl+2 Apply.EQMatchTrain
Match
 IDENTIFYING AUDIO PROBLEMS 228

71. Identifying Audio Problems

As with medical diagnostics, the key to successful audio restoration lies in your ability to correctly
analyze the subject’s condition. This can be a life-long, never-ending quest, constantly honing the
ear to distinguish the noises and audio events that need to be corrected.

RX Processing Step Flowchart

To get started, it’s important to identify the problems with your file and select the tool(s) that will
give you the best results. We have created a process flowchart to help identify the right RX module
to use and what modules to consider using if you aren’t getting the results you are looking for.
IDENTIFYING AUDIO PROBLEMS 229
 IDENTIFYING AUDIO PROBLEMS 230

Using the Spectrogram Display

Let’s briefly look at how to examine your audio using the spectrogram and waveform display tools,
then consider how to identify audio problems using these displays.
The aim of any good visualization tool for audio repair and restoration is to provide you with more
information about an audible problem. This not only helps inform your editing decisions, but, in the
case of a spectrogram display, can provide new, exciting ways to edit audio, especially when used
in tandem with a waveform display.

Hum
Hum is usually the result of electrical noise somewhere in the recorded signal chain. It’s normally
heard as a low-frequency tone based at either 50 Hz or 60 Hz depending on where the recording
was made If you zoom in to the low frequencies, you’ll be able to see hum as a series of horizontal
lines, usually with a bright line at 50 Hz or 60 Hz and several less intense lines above it at
harmonics. See the example below:

De-hum works best when frequencies of the hum do not overlap with any useful transient signals.

Buzz
In some cases, electrical noise will extend up to higher frequencies and manifest itself as a
background buzz. See the example below:

Hum-removal tools usually focus on low-frequency hum, so when the harmonics extend to
frequencies above 400 Hz, the Spectral De-noise tool is often more effective at removing the
problem.
 IDENTIFYING AUDIO PROBLEMS 231

Hiss and other Broadband Noise

Unlike hum and buzz, broadband noise is spread throughout the frequency spectrum and isn’t
concentrated at specific frequencies. Tape hiss and noise from fans and air conditioners are good
examples of broadband noise. In a spectrogram display, broadband noise usually appears as
speckles that surround the program material. See the example below:

Clicks, Pops, & Short Impulse Noises

Clicks and pops are common on recordings made from vinyl, shellac and other grooved media,
but can also be introduced by digital errors, including recording into a DAW with improper buffer
settings, or making a bad audio edit that missed a zero crossing. Even mouth noises such as
tongue clicks and lip smacks fall into the clicks category. These short impulse noises appear in
a spectrogram as vertical lines. The louder the click or pop, the brighter the line will appear. The
example below shows clicks and pops appearing in an audio recording transferred from vinyl:

The De-click tool can recognize, isolate, and then reduce and remove clicks like these.

Clipping
Clipping is an all-too-common problem. It can occur when a loud signal distorts the input to an
audio interface, analog-to-digital converter, mixing console, field recorder, or other sound capture
device. A spectrogram is not particularly useful for identifying clipped audio—for this you’ll want to
work with a waveform display. As you’ll see in the image below, the clipping appears as “squared-
off” sections of the waveform.
 IDENTIFYING AUDIO PROBLEMS 232

You can zoom in on a waveform and see in detail where the waveform has been truncated
because of clipping.

The De-clip tool can intelligently redraw the waveform to where it might have naturally been
if the signal hadn’t clipped. Sometimes, brickwall limited audio will also appear “squared off”
when zoomed out, but this doesn’t necessarily mean it will sound as heavily distorted as clipped
waveforms that have been truncated. You can zoom in to see if the tops of individual waveforms
are clipped.

Intermittent Noises
Intermittent noises are different than hiss and hum—they may appear infrequently and may not be
consistent in pitch or duration. Common examples include coughs, sneezes, footsteps, car horns,
ringing cell phones, etc. The images below represent two different examples of these noises:
 IDENTIFYING AUDIO PROBLEMS 233

The Spectral Repair tool can help isolate these intermittent sounds, analyze the audio around them
and attenuate or replace them.

Gaps and Drop Outs

Sometimes a recording may have short sections of missing or corrupted audio. These are usually
very obvious to both the eye and the ear! See the example below:

Deleting the gap and then applying Spectral Repair to replace any missing audio can help fix these
problems.
 AUTHORIZATION 234

72. Authorization
The first time you open the RX 11 Audio Editor or an RX plug-in, the Authorization window will
appear.

Selecting Sign In will open the Product Portal if this application is installed on your computer. If
Product Portal is not installed you will be brought to the web page to download Product Portal.
Selecting Manual Authorization will prompt the manual Authorization window:
 AUTHORIZATION 235

The Authorization window allows you to:

• TRIAL: Select the Continue in Trial option to continue a Trial period evaluation prior to
purchasing.
• DEMO: Continue evaluating the product with Demo limitations (after the 10 day Trial period
ends).
• AUTHORIZE: Authorize the product with a serial number.

Trial Mode
Trial mode allows you to evaluate RX 11 over a 10 day trial period. The trial period begins when
you first open the RX 11 Audio Editor or an RX plug-in in a DAW/NLE. The Authorization window
will display the number of days remaining in your trial period. Click the Continue button to exit the
Authorization window.

Demo Mode
After your 10 day trial period expires, you have the option to operate RX 11 in Demo mode. To
continue evaluating RX 11 in demo mode, click the Demo button.
 AUTHORIZATION 236

DEMO MODE LIMITATIONS

All controls and user interaction will be disabled in the RX 11 audio editor and plug-ins
while in Demo mode. However, playback and automation will remain intact so your
existing sessions are not negatively affected.

Authorization Methods
To disable Trial or Demo mode, you must authorize the product with a valid serial number. We offer
three authorization methods for RX 11:
• Online Authorization: Authorize RX 11 on a computer online.
• Offline Authorization: Authorize RX 11 on a computer offline.
• iLok Authorization: Authorize RX 11 using iLok.

More Authorization Help

For information about Authorization, please visit our Support Portal
 LICENSE INFORMATION 237

73. License Information

Abseil
Copyright 2020 The Abseil Authors.
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

Anti-Grain Geometry
Version 2.4
Copyright (c) 2002-2005 Maxim Shemanarev (McSeem).
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
3. The name of the author may not be used to endorse or promote products derived from this
software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE AUTHOR ‘‘AS IS’’ AND ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN
NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
SUCH DAMAGE.

ARA
ARA Audio Random Access is a Celemony product.
 LICENSE INFORMATION 238

ARM_NEON_2_x86_SSE
Copyright (C) 2012-2020 Intel Corporation. All rights reserved.
IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
By downloading, copying, installing or using the software you agree to this license. If you do not
agree to this license, do not download, install, copy or use the software.
License Agreement
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• The name of the copyright holders may not be used to endorse or promote products derived
from this software without specific prior written permission.
This software is provided by the copyright holders and contributors “as is” and any express
or implied warranties, including, but not limited to, the implied warranties of merchantability
and fitness for a particular purpose are disclaimed. In no event shall the Intel Corporation or
contributors be liable for any direct, indirect, incidental, special, exemplary, or consequential
damages (including, but not limited to, procurement of substitute goods or services; loss of use,
data, or profits; or business interruption) however caused and on any theory of liability, whether in
contract, strict liability, or tort (including negligence or otherwise) arising in any way out of the use
of this software, even if advised of the possibility of such damage.

base64
v0.4.0
Copyright (c) 2005-2007, Nick Galbreath
Copyright (c) 2013-2019, Alfred Klomp
Copyright (c) 2015-2017, Wojciech Mula
Copyright (c) 2016-2017, Matthieu Darbois
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
 LICENSE INFORMATION 239

• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Better Enums
Version 0.11.1
Copyright (c) 2012-2016, Anton Bachin. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Boost
Boost Software License - Version 1.0 - August 17th, 2003
Permission is hereby granted, free of charge, to any person or organization obtaining a copy of
the software and accompanying documentation covered by this license (the “Software”) to use,
reproduce, display, distribute, execute, and transmit the Software, and to prepare derivative works
of the Software, and to permit third-parties to whom the Software is furnished to do so, all subject
to the following:
The copyright notices in the Software and this entire statement, including the above license grant,
this restriction and the following disclaimer, must be included in all copies of the Software, in whole
or in part, and all derivative works of the Software, unless such copies or derivative works are
solely in the form of machine-executable object code generated by a source language processor.
 LICENSE INFORMATION 240

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT SHALL THE
COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE FOR ANY
DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.

Bravura
Copyright © 2015, Steinberg Media Technologies GmbH http://www.steinberg.net/, with Reserved
Font Name “Bravura”.
This Font Software is licensed under the SIL Open Font License, Version 1.1. This license is copied
below, and is also available with a FAQ at: http://scripts.sil.org/OFL
SIL OPEN FONT LICENSE Version 1.1 - 26 February 2007
PREAMBLE
The goals of the Open Font License (OFL) are to stimulate worldwide development of collaborative
font projects, to support the font creation efforts of academic and linguistic communities, and to
provide a free and open framework in which fonts may be shared and improved in partnership with
others.
The OFL allows the licensed fonts to be used, studied, modified and redistributed freely as long
as they are not sold by themselves. The fonts, including any derivative works, can be bundled,
embedded, redistributed and/or sold with any software provided that any reserved names are not
used by derivative works. The fonts and derivatives, however, cannot be released under any other
type of license. The requirement for fonts to remain under this license does not apply to any
document created using the fonts or their derivatives.
DEFINITIONS
“Font Software” refers to the set of files released by the Copyright Holder(s) under this license and
clearly marked as such. This may include source files, build scripts and documentation.
“Reserved Font Name” refers to any names specified as such after the copyright statement(s).
“Original Version” refers to the collection of Font Software components as distributed by the
Copyright Holder(s).
“Modified Version” refers to any derivative made by adding to, deleting, or substituting – in part or
in whole – any of the components of the Original Version, by changing formats or by porting the
Font Software to a new environment.
“Author” refers to any designer, engineer, programmer, technical writer or other person who
contributed to the Font Software.
PERMISSION & CONDITIONS
Permission is hereby granted, free of charge, to any person obtaining a copy of the Font Software,
to use, study, copy, merge, embed, modify, redistribute, and sell modified and unmodified copies of
the Font Software, subject to the following conditions:
1. Neither the Font Software nor any of its individual components, in Original or Modified Versions,
may be sold by itself.
2. Original or Modified Versions of the Font Software may be bundled, redistributed and/or sold
with any software, provided that each copy contains the above copyright notice and this
license. These can be included either as stand-alone text files, human-readable headers or in
the appropriate machine-readable metadata fields within text or binary files as long as those
fields can be easily viewed by the user.
 LICENSE INFORMATION 241

3. No Modified Version of the Font Software may use the Reserved Font Name(s) unless explicit
written permission is granted by the corresponding Copyright Holder. This restriction only
applies to the primary font name as presented to the users.
4. The name(s) of the Copyright Holder(s) or the Author(s) of the Font Software shall not be
used to promote, endorse or advertise any Modified Version, except to acknowledge the
contribution(s) of the Copyright Holder(s) and the Author(s) or with their explicit written
permission.
5. The Font Software, modified or unmodified, in part or in whole, must be distributed entirely
under this license, and must not be distributed under any other license. The requirement for
fonts to remain under this license does not apply to any document created using the Font
Software.
TERMINATION
This license becomes null and void if any of the above conditions are not met.
DISCLAIMER
THE FONT SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF COPYRIGHT, PATENT, TRADEMARK,
OR OTHER RIGHT. IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE FOR ANY CLAIM,
DAMAGES OR OTHER LIABILITY, INCLUDING ANY GENERAL, SPECIAL, INDIRECT, INCIDENTAL, OR
CONSEQUENTIAL DAMAGES, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
ARISING FROM, OUT OF THE USE OR INABILITY TO USE THE FONT SOFTWARE OR FROM OTHER
DEALINGS IN THE FONT SOFTWARE.

C++ Rest SDK

Version 2.10.15
Main Library:
Copyright (c) 2014, Peter Thorson. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of the WebSocket++ Project nor the names of its contributors may be
used to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL PETER THORSON BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
THE POSSIBILITY OF SUCH DAMAGE.
 LICENSE INFORMATION 242

Bundled Libraries:
****** Base 64 Library (base64/base64.hpp) ******
base64.hpp is a repackaging of the base64.cpp and base64.h files into a single header
suitable for use as a header only library. This conversion was done by Peter Thorson
([email protected]) in 2012. All modifications to the code are redistributed under the
same license as the original, which is listed below.
base64.cpp and base64.h
Copyright (C) 2004-2008 René Nyffenegger
This source code is provided ‘as-is’, without any express or implied warranty. In no event will the
author be held liable for any damages arising from the use of this software.
Permission is granted to anyone to use this software for any purpose, including commercial
applications, and to alter it and redistribute it freely, subject to the following restrictions:
1. The origin of this source code must not be misrepresented; you must not claim that you wrote
the original source code. If you use this source code in a product, an acknowledgment in the
product documentation would be appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be misrepresented as
being the original source code.
3. This notice may not be removed or altered from any source distribution.
René Nyffenegger [email protected]
****** SHA1 Library (sha1/sha1.hpp) ******
sha1.hpp is a repackaging of the sha1.cpp and sha1.h files from the shallsha1 library (http://
code.google.com/p/smallsha1/ ) into a single header suitable for use as a header only library. This
conversion was done by Peter Thorson ( [email protected]) in 2013. All modifications to
the code are redistributed under the same license as the original, which is listed below.
Copyright (c) 2011, Micael Hildenborg
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of Micael Hildenborg nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY Micael Hildenborg ‘‘AS IS’’ AND ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
EVENT SHALL Micael Hildenborg BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
SUCH DAMAGE.
 LICENSE INFORMATION 243

****** MD5 Library (common/md5.hpp) ******

md5.hpp is a reformulation of the md5.h and md5.c code from http://
www.opensource.apple.com/source/cups/cups-59/cups/md5.c to allow it to function as
a component of a header only library. This conversion was done by Peter Thorson
( [email protected]) in 2012 for the WebSocket++ project. The changes are released
under the same license as the original (listed below)
Copyright (C) 1999, 2002 Aladdin Enterprises. All rights reserved.
This software is provided ‘as-is’, without any express or implied warranty. In no event will the
authors be held liable for any damages arising from the use of this software.
Permission is granted to anyone to use this software for any purpose, including commercial
applications, and to alter it and redistribute it freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must not claim that you wrote
the original software. If you use this software in a product, an acknowledgment in the product
documentation would be appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be misrepresented as
being the original software.
3. This notice may not be removed or altered from any source distribution.
L. Peter Deutsch
[email protected]
****** UTF8 Validation logic (utf8_validation.hpp) ******
utf8_validation.hpp is adapted from code originally written by Bjoern Hoehrmann
[email protected] . See http://bjoern.hoehrmann.de/utf-8/decoder/dfa/ for details.
The original license:
Copyright (c) 2008-2009 Bjoern Hoehrmann [email protected]
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Caffe
COPYRIGHT
All contributions by the University of California:
Copyright (c) 2014-2017 The Regents of the University of California (Regents)
All rights reserved.
All other contributions:
 LICENSE INFORMATION 244

Copyright (c) 2014-2017, the respective contributors

All rights reserved.
Caffe uses a shared copyright model: each contributor holds copyright over their contributions to
Caffe. The project versioning records all such contribution and copyright details. If a contributor
wants to further mark their specific copyright on a particular contribution, they should indicate their
copyright solely in the commit message of the change when it is committed.
LICENSE
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
CONTRIBUTION AGREEMENT
By contributing to the BVLC/caffe repository through pull-request, comment, or otherwise, the
contributor releases their content to the license and copyright terms herein.

cerberus
Cerberus is a lightweight and extensible data validation library for Python.
ISC License
Copyright (c) 2012-2016 Nicola Iarocci.
Permission to use, copy, modify, and/or distribute this software for any purpose with or without fee
is hereby granted, provided that the above copyright notice and this permission notice appear in all
copies.
THE SOFTWARE IS PROVIDED “AS IS” AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY
AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM
LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE
OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
PERFORMANCE OF THIS SOFTWARE.

composable_kernel
https://github.com/ROCmSoftwarePlatform/composable_kernel
 LICENSE INFORMATION 245

Copyright (c) 2018- , Advanced Micro Devices, Inc. (Chao Liu, Jing Zhang)
Copyright (c) 2019- , Advanced Micro Devices, Inc. (Letao Qin, Qianfeng Zhang, Liang Huang,
Shaojie Wang)
Copyright (c) 2022- , Advanced Micro Devices, Inc. (Anthony Chang, Chunyu Lai, Illia Silin, Adam
Osewski, Poyen Chen, Jehandad Khan)
Copyright (c) 2019-2021, Advanced Micro Devices, Inc. (Hanwen Chang)
Copyright (c) 2019-2020, Advanced Micro Devices, Inc. (Tejash Shah)
Copyright (c) 2020 , Advanced Micro Devices, Inc. (Xiaoyan Zhou)
Copyright (c) 2021-2022, Advanced Micro Devices, Inc. (Jianfeng Yan)
SPDX-License-Identifier: MIT
Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

coremltools
BSD-3-Clause License
https://github.com/apple/coremltools
Copyright (c) 2020, Apple Inc. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
3. Neither the name of the copyright holder(s) nor the names of any contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
 LICENSE INFORMATION 246

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS

IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
© 2021 GitHub, Inc.

CpuInfo
Copyright (c) 2019 Google LLC
Copyright (c) 2017-2018 Facebook Inc.
Copyright (c) 2012-2017 Georgia Institute of Technology
Copyright (c) 2010-2012 Marat Dukhan
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Crashpad
Version 0.8.0
Copyright 2014 The Crashpad Authors. All rights reserved.
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at:
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.
 LICENSE INFORMATION 247

curl
https://github.com/curl
COPYRIGHT AND PERMISSION NOTICE
Copyright (C) Daniel Stenberg, [email protected], and many contributors, see the THANKS file.
All rights reserved.
Permission to use, copy, modify, and distribute this software for any purpose with or without fee
is hereby granted, provided that the above copyright notice and this permission notice appear in all
copies.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY RIGHTS. IN NO EVENT
SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
Except as contained in this notice, the name of a copyright holder shall not be used in advertising
or otherwise to promote the sale, use or other dealings in this Software without prior written
authorization of the copyright holder.

dcleblanc/SafeInt
MIT License
Copyright (c) 2018 Microsoft
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Distributed Machine Learning Common Codebase

Copyright (c) 2015 by Contributors
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.
 LICENSE INFORMATION 248

dlfcn-win32
dlfcn-win32 is licensed under the MIT License.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

DLPack
https://github.com/dmlc/dlpack
Copyright 2017 by Contributors
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

DLPack: Open In Memory Tensor Structure

Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

Eigen
Distributed under the Mozilla Public License v2.0 (MPLv2.0).
Full text of the license is available here: https://www.mozilla.org/en-US/MPL/2.0/
To receive a copy of the source code for the Eigen library distributed with this product under the
under the terms of the MPLv2.0 please contact [email protected].
 LICENSE INFORMATION 249

emsdk
https://github.com/emscripten-core/emsdk/
Copyright (c) 2018 Emscripten authors (see AUTHORS in Emscripten)
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
This is the MIT/Expat Licence. For more information see:
1. http://www.opensource.org/licenses/mit-license.php
2. http://en.wikipedia.org/wiki/MIT_License

FarmHash
Copyright (c) 2014 Google, Inc.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

FLAC
libFLAC and libFLAC++
Version 1.3.2
Copyright (c) 2000-2009 Josh Coalson
Copyright (c) 2011-2016 Xiph.Org Foundation
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
 LICENSE INFORMATION 250

• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of the Xiph.org Foundation nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

FlashAttention
FlashAttention, https://github.com/Dao-AILab/flash-attention
BSD 3-Clause License
Copyright (c) 2022, the respective contributors, as shown by the AUTHORS file. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of the copyright holder nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

FlatBuffers
Copyright 2014 Google Inc. All rights reserved.
 LICENSE INFORMATION 251

Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

FreeBSD: getopt.c file

Copyright (c) 1987, 1993, 1994
The Regents of the University of California.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
3. Neither the name of the University nor the names of its contributors may be used to endorse or
promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS’’ AND ANY
EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

FreeType
Version 2.4.6
Portions of this software are copyright © 2011 The FreeType Project (www.freetype.org). All rights
reserved.

FXDev
The MIT License (MIT)
Copyright (c) 2017 Facebook Inc.
Copyright (c) 2016-2017 Marat Dukhan
 LICENSE INFORMATION 252

Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

G3log: Asynchronous logger with Dynamic Sinks

This is free and unencumbered software released into the public domain.
Anyone is free to copy, modify, publish, use, compile, sell, or distribute this software, either in
source code form or as a compiled binary, for any purpose, commercial or non-commercial, and by
any means.
In jurisdictions that recognize copyright laws, the author or authors of this software dedicate any
and all copyright interest in the software to the public domain. We make this dedication for the
benefit of the public at large and to the detriment of our heirs and successors. We intend this
dedication to be an overt act of relinquishment in perpetuity of all present and future rights to this
software under copyright law.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS
BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
For more information, please refer to http://unlicense.org/

gemmlowp
Copyright 2021 gemmlowp authors
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

GLEW
The OpenGL Extension Wrangler Library
Copyright (C) 2002-2008, Milan Ikits <milan ikits[]ieee org>
 LICENSE INFORMATION 253

Copyright (C) 2002-2008, Marcelo E. Magallon <mmagallo[]debian org>

Copyright (C) 2002, Lev Povalahev
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• The name of the author may be used to endorse or promote products derived from this
software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Mesa 3-D graphics library Version: 7.0
Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL BRIAN PAUL
BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Copyright (c) 2007 The Khronos Group Inc.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and/or associated documentation files (the “Materials”), to deal in the Materials without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Materials, and to permit persons to whom the Materials are furnished to
do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Materials.
 LICENSE INFORMATION 254

THE MATERIALS ARE PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.

google/glog
Copyright (c) 2008, Google Inc.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of Google Inc. nor the names of its contributors may be used to endorse or
promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
A function gettimeofday in utilities.cc is based on
http://www.google.com/codesearch/p?hl=en#dR3YEbitojA/
COPYING&q=GetSystemTimeAsFileTime%20license:bsd
The license of this code is:
Copyright (c) 2003-2008, Jouni Malinen [email protected] and contributors
All Rights Reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
3. Neither the name(s) of the above-listed copyright holder(s) nor the names of its contributors
may be used to endorse or promote products derived from this software without specific prior
written permission.
 LICENSE INFORMATION 255

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS

IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

google/nsync
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

google/re2
Copyright (c) 2009 The RE2 Authors. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of Google Inc. nor the names of its contributors may be used to endorse or
promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

google/sentencepiece
https://github.com/google/sentencepiece
(included when statically linked with onnxruntime-extensions)
 LICENSE INFORMATION 256

Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

google/XNNPACK
BSD License
For XNNPACK software
Copyright © Facebook, Inc. and its affiliates. All rights reserved.
Copyright 2019 Google LLC
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name Facebook nor the names of its contributors may be used to endorse or
promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

gsl
Copyright (c) 2015 Microsoft Corporation. All rights reserved.
This code is licensed under the MIT License (MIT).
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
 LICENSE INFORMATION 257

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

HalidelR
Copyright (c) 2016 HalideIR contributors
Copyright (c) 2012-2014 MIT CSAIL, Google Inc., and other contributors
HalideIR is derived from the Halide project.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

HowardHinnant/date
The source code in this project is released using the MIT License. There is no global license for the
project because each file is licensed individually with different author names and/or dates.
If you contribute to this project, please add your name to the license of each file you modify. If you
have already contributed to this project and forgot to add your name to the license, please feel free
to submit a new P/R to add your name to the license in each file you modified.
For convenience, here is a copy of the MIT license found in each file except without author names
or dates:
The MIT License (MIT)
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
 LICENSE INFORMATION 258

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

huggingface/transformers
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

IcoMoon
IcoMoon-Free licensed under: CC BY 4.0

Intel Math Kernel Library (Intel MKL)

Intel Simplified Software License (Version April 2018)
Copyright (c) 2018 Intel Corporation.
Use and Redistribution.
You may use and redistribute the software (the “Software”), without modification, provided the
following conditions are met:
• Redistributions must reproduce the above copyright notice and the following terms of use in
the Software and in the documentation and/or other materials provided with the distribution.
• Neither the name of Intel nor the names of its suppliers may be used to endorse or promote
products derived from this Software without specific prior written permission.
• No reverse engineering, decompilation, or disassembly of this Software is permitted.
Limited patent license. Intel grants you a world-wide, royalty-free, non-exclusive license under
patents it now or hereafter owns or controls to make, have made, use, import, offer to sell and
sell (“Utilize”) this Software, but solely to the extent that any such patent is necessary to Utilize
the Software alone. The patent license shall not apply to any combinations which include this
software. No hardware per se is licensed hereunder.
Third party and other Intel programs. “Third Party Programs” are the files listed in the “third-
party-programs.txt” text file that is included with the Software and may include Intel programs
under separate license terms. Third Party Programs, even if included with the distribution of the
Materials, are governed by separate license terms and those license terms solely govern your use
of those programs.
 LICENSE INFORMATION 259

DISCLAIMER. THIS SOFTWARE IS PROVIDED “AS IS” AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT ARE
DISCLAIMED. THIS SOFTWARE IS NOT INTENDED FOR USE IN SYSTEMS OR APPLICATIONS
WHERE FAILURE OF THE SOFTWARE MAY CAUSE PERSONAL INJURY OR DEATH AND YOU
AGREE THAT YOU ARE FULLY RESPONSIBLE FOR ANY CLAIMS, COSTS, DAMAGES, EXPENSES,
AND ATTORNEYS’ FEES ARISING OUT OF ANY SUCH USE, EVEN IF ANY CLAIM ALLEGES THAT
INTEL WAS NEGLIGENT REGARDING THE DESIGN OR MANUFACTURE OF THE MATERIALS.
LIMITATION OF LIABILITY. IN NO EVENT WILL INTEL BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
THE POSSIBILITY OF SUCH DAMAGE. YOU AGREE TO INDEMNIFY AND HOLD INTEL HARMLESS
AGAINST ANY CLAIMS AND EXPENSES RESULTING FROM YOUR USE OR UNAUTHORIZED USE
OF THE SOFTWARE.
No support. Intel may make changes to the Software, at any time without notice, and is not
obligated to support, update or provide training for the Software.
Termination. Intel may terminate your right to use the Software in the event of your breach of this
Agreement and you fail to cure the breach within a reasonable period of time.
Feedback. Should you provide Intel with comments, modifications, corrections, enhancements
or other input (“Feedback”) related to the Software Intel will be free to use, disclose, reproduce,
license or otherwise distribute or exploit the Feedback in its sole discretion without any obligations
or restrictions of any kind, including without limitation, intellectual property rights or licensing
obligations.
Compliance with laws. You agree to comply with all relevant laws and regulations governing your
use, transfer, import or export (or prohibition thereof) of the Software.
Governing law. All disputes will be governed by the laws of the United States of America and
the State of Delaware without reference to conflict of law principles and subject to the exclusive
jurisdiction of the state or federal courts sitting in the State of Delaware, and each party agrees
that it submits to the personal jurisdiction and venue of those courts and waives any objections.
The United Nations Convention on Contracts for the International Sale of Goods (1980) is
specifically excluded and will not apply to the Software.
*Other names and brands may be claimed as the property of others.

Intel neural-compressor
https://github.com/intel/neural-compressor
Copyright 2016-2019 Intel Corporation
Copyright 2018 YANDEX LLC
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.
 LICENSE INFORMATION 260

intel/dnnl
Copyright 2016-2018 Intel Corporation
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.
sub-components:
xbyak
Copyright (c) 2007 MITSUNARI Shigeo. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and
the following disclaimer. Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation and/or other materials
provided with the distribution. Neither the name of the copyright owner nor the names of its
contributors may be used to endorse or promote products derived from this software without
specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Intel® Integrated Performance Primitives (Intel® IPP)

Version 2019.0.5
LIMITATION OF LIABILITY. IN NO EVENT WILL INTEL BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
THE POSSIBILITY OF SUCH DAMAGE. YOU AGREE TO INDEMNIFY AND HOLD INTEL HARMLESS
AGAINST ANY CLAIMS AND EXPENSES RESULTING FROM YOUR USE OR UNAUTHORIZED USE
OF THE SOFTWARE.
No support. Intel may make changes to the Software, at any time without notice, and is not
obligated to support, update or provide training for the Software.
Termination. Intel may terminate your right to use the Software in the event of your breach of this
Agreement and you fail to cure the breach within a reasonable period of time.
 LICENSE INFORMATION 261

Feedback. Should you provide Intel with comments, modifications, corrections, enhancements
or other input (“Feedback”) related to the Software Intel will be free to use, disclose, reproduce,
license or otherwise distribute or exploit the Feedback in its sole discretion without any obligations
or restrictions of any kind, including without limitation, intellectual property rights or licensing
obligations.
Compliance with laws. You agree to comply with all relevant laws and regulations governing your
use, transfer, import or export (or prohibition thereof) of the Software.
Governing law. All disputes will be governed by the laws of the United States of America and
the State of Delaware without reference to conflict of law principles and subject to the exclusive
jurisdiction of the state or federal courts sitting in the State of Delaware, and each party agrees
that it submits to the personal jurisdiction and venue of those courts and waives any objections.
The United Nations Convention on Contracts for the International Sale of Goods (1980) is
specifically excluded and will not apply to the Software.
Other names and brands may be claimed as the property of others.

JDAI-CV/DNNLibrary
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

JsonCpp
Version 1.2.1
Copyright (c) 2007-2010 Baptiste Lepilleur and The JsonCpp Authors
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

LAME
libmp3lame
Version 3.99.5
Copyright (c) 1999-2011 The L.A.M.E. Team
 LICENSE INFORMATION 262

Distributed under the GNU Library General Public License v2 (LGPLv2).

Full text of the license is available here: https://www.gnu.org/licenses/old-licenses/lgpl-2.0.txt
To receive a copy of the source code for the LAME library distributed with this product under the
under the terms of the LGPLv2 please contact [email protected].

libb64
https://github.com/libb64/libb64
Copyright-Only Dedication (based on United States law) or Public Domain Certification
The person or persons who have associated work with this document (the “Dedicator” or “Certifier”)
hereby either (a) certifies that, to the best of his knowledge, the work of authorship identified is
in the public domain of the country from which the work is published, or (b) hereby dedicates
whatever copyright the dedicators holds in the work of authorship identified below (the “Work”)
to the public domain. A certifier, moreover, dedicates any copyright interest he may have in the
associated work, and for these purposes, is described as a “dedicator” below.
A certifier has taken reasonable steps to verify the copyright status of this work. Certifier
recognizes that his good faith efforts may not shield him from liability if in fact the work certified is
not in the public domain.
Dedicator makes this dedication for the benefit of the public at large and to the detriment of
the Dedicator’s heirs and successors. Dedicator intends this dedication to be an overt act of
relinquishment in perpetuity of all present and future rights under copyright law, whether vested or
contingent, in the Work. Dedicator understands that such relinquishment of all rights includes the
relinquishment of all rights to enforce (by lawsuit or otherwise) those copyrights in the Work.
Dedicator recognizes that, once placed in the public domain, the Work may be freely reproduced,
distributed, transmitted, used, modified, built upon, or otherwise exploited by anyone for any
purpose, commercial or non-commercial, and in any way, including by methods that have not
yet been invented or conceived.

libprotobuf-mutator
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

LibXML2
Version 2.7.8
Except where otherwise noted in the source code (e.g. the files hash.c, list.c and the trio files, which
are covered by a similar licence but with different Copyright notices) all the files are:
Copyright (C) 1998-2003 Daniel Veillard. All Rights Reserved.
 LICENSE INFORMATION 263

Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE DANIEL
VEILLARD BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Except as contained in this notice, the name of Daniel Veillard shall not be used in advertising
or otherwise to promote the sale, use or other dealings in this Software without prior written
authorization from him.

madler/zlib
The deflate format used by zlib was defined by Phil Katz. The deflate and zlib specifications
were written by L. Peter Deutsch. Thanks to all the people who reported problems and suggested
various improvements in zlib; they are too numerous to cite here.
Copyright notice:
(C) 1995-2017 Jean-loup Gailly and Mark Adler
This software is provided ‘as-is’, without any express or implied warranty. In no event will the
authors be held liable for any damages arising from the use of this software.
Permission is granted to anyone to use this software for any purpose, including commercial
applications, and to alter it and redistribute it freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must not claim that you wrote
the original software. If you use this software in a product, an acknowledgment in the product
documentation would be appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be misrepresented as
being the original software.
3. This notice may not be removed or altered from any source distribution.
Jean-loup Gailly Mark Adler [email protected] [email protected]
If you use the zlib library in a product, we would appreciate not receiving lengthy legal documents
to sign. The sources are provided for free but without warranty of any kind. The library has been
entirely written by Jean-loup Gailly and Mark Adler; it does not include third-party code.
If you redistribute modified sources, we would appreciate that you include in the file ChangeLog
history information documenting your changes. Please read the FAQ for more information on the
distribution of modified source versions.

magic_enum
MIT License
Copyright (c) 2019 - 2023 Daniil Goncharov
 LICENSE INFORMATION 264

Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Microsoft Cognitive Toolkit (CNTK)

Copyright (c) Microsoft Corporation. All rights reserved.
MIT License
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

microsoft/mimalloc
microsoft/mimalloc, https://github.com/microsoft/mimalloc
MIT License
Copyright (c) 2018-2021 Microsoft Corporation, Daan Leijen
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
 LICENSE INFORMATION 265

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

microsoft/wil
MIT License
Copyright (c) Microsoft Corporation. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE

mpi4py
https://github.com/mpi4py/mpi4py/
=====================
LICENSE: MPI for Python
=====================
:Author: Lisandro Dalcin
:Contact: [email protected]
Copyright (c) 2019, Lisandro Dalcin.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
 LICENSE INFORMATION 266

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS “AS IS”
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

MurmurHash3
MIT license
https://github.com/aappleby/smhasher
SMHasher is a test suite designed to test the distribution, collision, and performance properties of
non-cryptographic hash functions.
This is the home for the MurmurHash family of hash functions along with the SMHasher test suite
used to verify them.
SMHasher is released under the MIT license. All MurmurHash versions are public domain software,
and the author disclaims all copyright to their code.

Netlib numeralgo na10 Aberth’s method

All the software contained in this library is protected by copyright. Permission to use, copy, modify,
and distribute this software for any purpose without fee is hereby granted, provided that this entire
notice is included in all copies of any software which is or includes a copy or modification of this
software and in all copies of the supporting documentation for such software.
THIS SOFTWARE IS BEING PROVIDED “AS IS”, WITHOUT ANY EXPRESS OR IMPLIED WARRANTY.
IN NO EVENT, NEITHER THE AUTHORS, NOR THE PUBLISHER, NOR ANY MEMBER OF THE
EDITORIAL BOARD OF THE JOURNAL “NUMERICAL ALGORITHMS”, NOR ITS EDITOR-IN-CHIEF, BE
LIABLE FOR ANY ERROR IN THE SOFTWARE, ANY MISUSE OF IT OR ANY DAMAGE ARISING OUT
OF ITS USE. THE ENTIRE RISK OF USING THE SOFTWARE LIES WITH THE PARTY DOING SO.
ANY USE OF THE SOFTWARE CONSTITUTES ACCEPTANCE OF THE TERMS OF THE ABOVE
STATEMENT.
AUTHOR:
• DARIO ANDREA
UNIVERSITY OF PISA, ITALY
E-MAIL: [email protected]
REFERENCE:
• NUMERICAL COMPUTATION OF POLYNOMIAL ZEROS BY MEANS OF ABERTH’S METHOD
NUMERICAL ALGORITHMS, 13 (1996), PP. 179-200
SOFTWARE REVISION DATE:
• JUNE, 1996
SOFTWARE LANGUAGE:
• FORTRAN
 LICENSE INFORMATION 267

nlohmann/json
MIT License
Copyright (c) 2013-2019 Niels Lohmann
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

nvidia/cutlass
Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
SPDX-License-Identifier: BSD-3-Clause
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
3. Neither the name of the copyright holder nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

OGG / Vorbis
libogg and libvorbis
Version 1.3.2
Copyright (c) 2014, Xiph.org Foundation
 LICENSE INFORMATION 268

Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/ or other materials provided with the
distribution.
• Neither the name of the Xiph.org Foundation nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

onnx
Open Neural Network Exchange
Copyright (c) Facebook, Inc. and Microsoft Corporation. All rights reserved.
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

onnx/onnx-tensorrt
MIT License
Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
Copyright (c) 2018 Open Neural Network Exchange
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
 LICENSE INFORMATION 269

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

onnxruntime
MIT License
Copyright (c) Microsoft Corporation
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
THIRD PARTY SOFTWARE NOTICES AND INFORMATION
Do Not Translate or Localize
This software incorporates material from third parties. Microsoft makes certain open source code
available at http://3rdpartysource.microsoft.com, or you may send a check or money order for US
$5.00, including the product name, the open source component name, and version number, to:
Source Code Compliance Team
Microsoft Corporation
One Microsoft Way
Redmond, WA 98052
USA
Notwithstanding any other terms, you may reverse engineer this software to the extent required to
debug changes to any libraries licensed under the GNU Lesser General Public License.

Open MPI
3-Clause BSD License
Most files in this release are marked with the copyrights of the organizations who have edited
them. The copyrights below are in no particular order and generally reflect members of the Open
MPI core team who have contributed code to this release. The copyrights for code used under
license from other parties are included in the corresponding files.
Copyright (c) 2004-2010 The Trustees of Indiana University and Indiana University Research and
Technology Corporation. All rights reserved.
 LICENSE INFORMATION 270

Copyright (c) 2004-2017 The University of Tennessee and The University of Tennessee Research
Foundation. All rights reserved.
Copyright (c) 2004-2010 High Performance Computing Center Stuttgart, University of Stuttgart. All
rights reserved.
Copyright (c) 2004-2008 The Regents of the University of California. All rights reserved.
Copyright (c) 2006-2017 Los Alamos National Security, LLC. All rights reserved.
Copyright (c) 2006-2017 Cisco Systems, Inc. All rights reserved.
Copyright (c) 2006-2010 Voltaire, Inc. All rights reserved.
Copyright (c) 2006-2017 Sandia National Laboratories. All rights reserved.
Copyright (c) 2006-2010 Sun Microsystems, Inc. All rights reserved. Use is subject to license terms.
Copyright (c) 2006-2017 The University of Houston. All rights reserved.
Copyright (c) 2006-2009 Myricom, Inc. All rights reserved.
Copyright (c) 2007-2017 UT-Battelle, LLC. All rights reserved.
Copyright (c) 2007-2017 IBM Corporation. All rights reserved.
Copyright (c) 1998-2005 Forschungszentrum Juelich, Juelich Supercomputing Centre, Federal
Republic of Germany
Copyright (c) 2005-2008 ZIH, TU Dresden, Federal Republic of Germany
Copyright (c) 2007 Evergrid, Inc. All rights reserved.
Copyright (c) 2008 Chelsio, Inc. All rights reserved.
Copyright (c) 2008-2009 Institut National de Recherche en Informatique. All rights reserved.
Copyright (c) 2007 Lawrence Livermore National Security, LLC. All rights reserved.
Copyright (c) 2007-2017 Mellanox Technologies. All rights reserved.
Copyright (c) 2006-2010 QLogic Corporation. All rights reserved.
Copyright (c) 2008-2017 Oak Ridge National Labs. All rights reserved.
Copyright (c) 2006-2012 Oracle and/or its affiliates. All rights reserved.
Copyright (c) 2009-2015 Bull SAS. All rights reserved.
Copyright (c) 2010 ARM ltd. All rights reserved.
Copyright (c) 2016 ARM, Inc. All rights reserved.
Copyright (c) 2010-2011 Alex Brick. All rights reserved.
Copyright (c) 2012 The University of Wisconsin-La Crosse. All rights reserved.
Copyright (c) 2013-2016 Intel, Inc. All rights reserved.
Copyright (c) 2011-2017 NVIDIA Corporation. All rights reserved.
Copyright (c) 2016 Broadcom Limited. All rights reserved.
Copyright (c) 2011-2017 Fujitsu Limited. All rights reserved.
Copyright (c) 2014-2015 Hewlett-Packard Development Company, LP. All rights reserved.
Copyright (c) 2013-2017 Research Organization for Information Science (RIST). All rights reserved.
Copyright (c) 2017-2018 Amazon.com, Inc. or its affiliates. All Rights reserved.
Copyright (c) 2018 DataDirect Networks. All rights reserved.
Copyright (c) 2018-2019 Triad National Security, LLC. All rights reserved.
 LICENSE INFORMATION 271

Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer listed in this license in the documentation and/or other materials
provided with the distribution.
• Neither the name of the copyright holders nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
The copyright holders provide no reassurances that the source code provided does not infringe
any patent, copyright, or any other intellectual property rights of third parties. The copyright holders
disclaim any liability to any recipient for claims brought against recipient by any third party for
infringement of that parties intellectual property rights.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Opus
Copyright 2001-2011 Xiph.Org, Skype Limited, Octasic,
Jean-Marc Valin, Timothy B. Terriberry,
CSIRO, Gregory Maxwell, Mark Borgerding,
Erik de Castro Lopo
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of Internet Society, IETF or IETF Trust, nor the names of specific contributors,
may be used to endorse or promote products derived from this software without specific prior
written permission.
 LICENSE INFORMATION 272

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS

IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Opus is subject to the royalty-free patent licenses which are specified at:
Xiph.Org Foundation:
https://datatracker.ietf.org/ipr/1524/
Microsoft Corporation:
https://datatracker.ietf.org/ipr/1914/
Broadcom Corporation:
https://datatracker.ietf.org/ipr/1526/

PIL
The Python Imaging Library (PIL) is
Copyright © 1997-2011 by Secret Labs AB
Copyright © 1995-2011 by Fredrik Lundh
Pillow is the friendly PIL fork. It is
Copyright © 2010-2023 by Alex Clark and contributors
Like PIL, Pillow is licensed under the open source HPND License:
By obtaining, using, and/or copying this software and/or its associated documentation, you agree
that you have read, understood, and will comply with the following terms and conditions:
Permission to use, copy, modify, and distribute this software and its associated documentation for
any purpose and without fee is hereby granted, provided that the above copyright notice appears
in all copies, and that both that copyright notice and this permission notice appear in supporting
documentation, and that the name of Secret Labs AB or the author not be used in advertising or
publicity pertaining to distribution of the software without specific, written prior permission.
SECRET LABS AB AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS
SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO
EVENT SHALL SECRET LABS AB OR THE AUTHOR BE LIABLE FOR ANY SPECIAL, INDIRECT OR
CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
SOFTWARE.

portmidi
PortMidi Portable Real-Time MIDI Library
Copyright (c) 1999-2000 Ross Bencina and Phil Burk
Copyright (c) 2001-2009 Roger B. Dannenberg
 LICENSE INFORMATION 273

Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

posix pthread library

https://sourceforge.net/projects/pthreads4w
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

protocolbuffers/protobuf
Copyright 2008 Google Inc. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of Google Inc. nor the names of its contributors may be used to endorse or
promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 LICENSE INFORMATION 274

Code generated by the Protocol Buffer compiler is owned by the owner of the input file used when
generating it. This code is not standalone and requires a support library to be linked with it. This
support library is itself covered by the above license.

pthreadpool
Copyright 2019 Google LLC
Copyright (c) 2017 Facebook Inc.
Copyright (c) 2015-2017 Georgia Institute of Technology
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Pytorch / Caffe2
From PyTorch:
Copyright (c) 2016- Facebook, Inc (Adam Paszke)
Copyright (c) 2014- Facebook, Inc (Soumith Chintala)
Copyright (c) 2011-2014 Idiap Research Institute (Ronan Collobert)
Copyright (c) 2012-2014 Deepmind Technologies (Koray Kavukcuoglu)
Copyright (c) 2011-2012 NEC Laboratories America (Koray Kavukcuoglu)
Copyright (c) 2011-2013 NYU (Clement Farabet)
Copyright (c) 2006-2010 NEC Laboratories America (Ronan Collobert, Leon Bottou, Iain Melvin,
Jason Weston)
Copyright (c) 2006 Idiap Research Institute (Samy Bengio)
Copyright (c) 2001-2004 Idiap Research Institute (Ronan Collobert, Samy Bengio, Johnny
Mariethoz)
From Caffe2:
Copyright (c) 2016-present, Facebook Inc. All rights reserved.
All contributions by Facebook:
 LICENSE INFORMATION 275

Copyright (c) 2016 Facebook Inc.

All contributions by Google:
Copyright (c) 2015 Google Inc.
All rights reserved.
All contributions by Yangqing Jia:
Copyright (c) 2015 Yangqing Jia
All rights reserved.
All contributions from Caffe:
Copyright(c) 2013, 2014, 2015, the respective contributors
All rights reserved.
All other contributions:
Copyright(c) 2015, 2016 the respective contributors
All rights reserved.
Caffe2 uses a copyright model similar to Caffe: each contributor holds copyright over their
contributions to Caffe2. The project versioning records all such contribution and copyright details.
If a contributor wants to further mark their specific copyright on a particular contribution, they
should indicate their copyright solely in the commit message of the change when it is committed.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
3. Neither the names of Facebook, Deepmind Technologies, NYU, NEC Laboratories America and
IDIAP Research Institute nor the names of its contributors may be used to endorse or promote
products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

pytorch/cpuinfo
BSD 2-Clause “Simplified” License
https://github.com/pytorch/cpuinfo
Copyright © 2019 Google LLC
Copyright © 2017-2018 Facebook Inc.
 LICENSE INFORMATION 276

Copyright © 2012-2017 Georgia Institute of Technology

Copyright © 2010-2012 Marat Dukhan
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

react-native
MIT License
https://github.com/facebook/react-native
Copyright © Facebook, Inc. and its affiliates.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

readerwriterqueue
Copyright (c) 2013-2015, Cameron Desrochers All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
 LICENSE INFORMATION 277

• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Resemblyzer
Copyright 2019 Resemblyzer Authors.
All rights reserved.
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at:
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

Roboto font family

Font data Copyright Google 2012
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at:
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

ruy
Copyright 2020 Google LLC. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.
 LICENSE INFORMATION 278

Skia
Copyright (c) 2011 Google Inc. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution. Neither the name of Google Inc. nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

SQLite
SQLite is Public Domain
All of the code and documentation in SQLite has been dedicated to the public domain by the
authors. All code authors, and representatives of the companies they work for, have signed
affidavits dedicating their contributions to the public domain and originals of those signed
affidavits are stored in a firesafe at the main offices of Hwaci. Anyone is free to copy, modify,
publish, use, compile, sell, or distribute the original SQLite code, either in source code form or as a
compiled binary, for any purpose, commercial or non-commercial, and by any means.
The previous paragraph applies to the deliverable code and documentation in SQLite - those parts
of the SQLite library that you actually bundle and ship with a larger application. Some scripts
used as part of the build process (for example the “configure” scripts generated by autoconf)
might fall under other open-source licenses. Nothing from these build scripts ever reaches the final
deliverable SQLite library, however, and so the licenses associated with those scripts should not be
a factor in assessing your rights to copy and use the SQLite library.
All of the deliverable code in SQLite has been written from scratch. No code has been taken from
other projects or from the open internet. Every line of code can be traced back to its original author,
and all of those authors have public domain dedications on file. So the SQLite code base is clean
and is uncontaminated with licensed code from other projects.

TagLib
Version 1.9.1
Copyright (c) 1995-2015 Scott Wheeler, Lukas Lalinsky, Ismael Orenstein, Allan Sandfeld Jensen,
Teemu Tervo, Mathias Panzenböck, and Tsuda Kageyu.
Distributed under the Mozilla Public License v1.1 (MPLv1.1).
 LICENSE INFORMATION 279

Full text of the license is available here: https://www.mozilla.org/en-US/MPL/1.1/

To receive a copy of the source code for the TagLib library distributed with this product under the
terms of the MPLv1.1 please contact [email protected].

Tencent/rapidjson
https://github.com/Tencent/rapidjson
Tencent is pleased to support the open source community by making RapidJSON available.
Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved.
If you have downloaded a copy of the RapidJSON binary from Tencent, please note that the
RapidJSON binary is licensed under the MIT License. If you have downloaded a copy of the
RapidJSON source code from Tencent, please note that RapidJSON source code is licensed under
the MIT License, except for the third-party components listed below which are subject to different
license terms. Your integration of RapidJSON into your own projects may require compliance with
the MIT License, as well as the other licenses applicable to the third-party components included
within RapidJSON. To avoid the problematic JSON license in your own projects, it’s sufficient to
exclude the bin/jsonchecker/ directory, as it’s the only code under the JSON license.
A copy of the MIT License is included in this file.
Other dependencies and licenses:
Open Source Software Licensed Under the BSD License:
The msinttypes r29
Copyright (c) 2006-2013 Alexander Chemeris
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of copyright holder nor the names of its contributors may be used to endorse
or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS’’ AND ANY
EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE REGENTS AND CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Open Source Software Licensed Under the JSON License:
json.org
Copyright (c) 2002 JSON.org
All Rights Reserved.
 LICENSE INFORMATION 280

JSON_checker
Copyright (c) 2002 JSON.org
All Rights Reserved.
Terms of the JSON License:
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
The Software shall be used for Good, not Evil.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Terms of the MIT License:
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Tensorflow
Copyright 2018 The TensorFlow Authors. All rights reserved.
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License
is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
either express or implied. See the License for the specific language governing permissions and
limitations under the License.

TinyXML
Copyright (c) 2000-2002 Lee Thomason (www.grinninglizard.com)
 LICENSE INFORMATION 281

This software is provided ‘as-is’, without any express or implied warranty. In no event will the
authors be held liable for any damages arising from the use of this software.
Permission is granted to anyone to use this software for any purpose, including commercial
applications, and to alter it and redistribute it freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must not claim that you wrote
the original software. If you use this software in a product, an acknowledgment in the product
documentation would be appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be misrepresented as
being the original software.
3. This notice may not be removed or altered from any source distribution.

Tipue Search
Copyright (c) 2017 Tipue
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

vectorize
Copyright (c) 2012 Aaron Wishnick. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• Neither the name of the <organization> nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
 LICENSE INFORMATION 282

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS

IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL <COPYRIGHT HOLDER> BE LIABLE FOR ANY DIRECT,
INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
THE POSSIBILITY OF SUCH DAMAGE.

WebView2
Copyright (C) Microsoft Corporation. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
• The name of Microsoft Corporation, or the names of its contributors may not be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

xsimd
Copyright (c) 2016, Johan Mabille, Sylvain Corlay, Wolf Vollprecht and Martin Renou
Copyright (c) 2016, QuantStack
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted
provided that the following conditions are met:
• Redistributions of source code must retain the above copyright notice, this list of conditions
and the following disclaimer.
• Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.
 LICENSE INFORMATION 283

• Neither the name of the copyright holder nor the names of its contributors may be used
to endorse or promote products derived from this software without specific prior written
permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS
IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Yoga
Version 1.9.0
MIT License
Copyright (c) 2014-present, Facebook, Inc.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software
and associated documentation files (the “Software”), to deal in the Software without restriction,
including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense,
and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do
so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

ZeroMQ
Copyright (c) 2007-2015 iMatix Corporation
Copyright (c) 2009-2011 250bpm s.r.o.
Copyright (c) 2010-2011 Miru Limited
Copyright (c) 2011 VMware, Inc.
Copyright (c) 2012 Spotify AB
Copyright (c) 2013 Ericsson AB
Copyright (c) 2014 AppDynamics Inc.
Distributed under the GNU Lesser General Public License v3 (LGPLv3) with the following “Static
linking exception”:
 LICENSE INFORMATION 284

The copyright holders give you permission to link this library with independent modules to produce
an executable, regardless of the license terms of these independent modules, and to copy and
distribute the resulting executable under terms of your choice, provided that you also meet, for
each linked independent module, the terms and conditions of the license of that module. An
independent module is a module which is not derived from or based on this library. If you modify
this library, you must extend this exception to your version of the library.
Full text of the license is available here: https://www.gnu.org/licenses/lgpl-3.0.txt
To receive a copy of the source code for the ZeroMQ library distributed with this product under the
under the terms of the LGPLv3 please contact [email protected].

zlib
Copyright (c) 1995-2004 Jean-loup Gailly and Mark Adler.
This software is provided ‘as-is’, without any express or implied warranty. In no event will the
authors be held liable for any damages arising from the use of this software.
Permission is granted to anyone to use this software for any purpose, including commercial
applications, and to alter it and redistribute it freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must not claim that you wrote
the original software. If you use this software in a product, an acknowledgment in the product
documentation would be appreciated but is not required.
2. Altered source versions must be plainly marked as such, and must not be misrepresented as
being the original software.
3. This notice may not be removed or altered from any source distribution.
Jean-loup Gailly [email protected]
Mark Adler [email protected]