Shree Ramdeobaba College Of Engineering And Management

Department of Electronics And Communication(2020-2021)

Subject: Digital Communication (ECT-401), (7th Sem, B section)

Teacher’s assessment-5
Topic: Code Excited Linear Predictive Coding.
(Group no.5)
Roll no. Name
44 Aditya Shailesh Namdeo
45 Rohit Vaishali Motdhare
51 Sakshi Manish Rathod
52 Sakshi Sanjay Baldawa
53 Saloni Sanjay Jais
54 Sana Kuresh Naseem
55 Sanskruti Ajay Deshpande
56 Himanshu Ravindra Channe
 INTRODUCTION
1. Inventers:
In 1985, Bishnu S. Atal and Manfred R. Schroeder proposed code excited linear prediction(CELP) which is a predictive
speech coding algorithm.
2. The CELP technique is based on three ideas:
* The use of a linear prediction (LP) model to model the vocal tract.
* The use of (adaptive and fixed) codebook entries as input (excitation) of the LP model.
* The search performed in closed-loop in a “perceptually weighted domain”.
3. CELP decoder :
The excitation is produced by summing the contributions from fixed and adaptive codebooks, where = fixed
codebook contribution ; = adaptive codebook contribution. The fixed codebook is a vector quantization dictionary
that is hard-coded into the codec. The entries in the adaptive codebook consist of delayed versions of the excitation. This
makes it possible to efficiently code periodic signals, such as voiced sounds.
4. CELP encoder :
The encoding (analysis) is performed by perceptually optimizing the decoded (synthesis) signal in a closed loop. The
CELP search is broken down into smaller, more manageable, sequential searches using a simple perceptual weighting
function for achieving real-time encoding.
 BRIEF EXPLAINATION
LPC Vocoder:
1. It has two key components: analysis or encoding and synthesis or decoding. The analysis part of LPC involves examining
the speech signal and breaking it down into segments or blocks. Each segment is than examined further to find the answer
to several key questions. Is the segment voiced or unvoiced ?, What is the pitch of the segment?, etc.
2. LPC analysis is usually conducted by a sender who answers these questions and usually transmits these answers onto a
receiver. The receiver performs LPC synthesis by using the answers received to build a filter that when provided the correct
input source will be able to accurately reproduce the original speech signal. Essentially LPC synthesis tries to imitate human
speech production.
3. In CELP, each trial waveform is synthesized by passing it through a two part cascade synthesis filter. The first part,
termed the pitch synthesis filter, inserts pitch periodicities into the reconstructed speech. The second filter is the formant
synthesis filter which introduces a frequency shaping related to the formant resonances produced by the human vocal tract.
Both filters are all-pole structures, using an FIR filter in a feedback configuration.
4. CELP is based on a source-filter model of speech, where linear prediction is used to model the filtering effect of the vocal tract (and
other effects) and this filter is excited by the speech source, viz. the glottal source and turbulent noise. Typically, the pitch or fundamental
frequency model is a long-term prediction (LTP) filter, which is just a linear predictor with a long delay
 Working
CELP Encoder CELP Decoder
For each of the excitation VQ codebook vectors, the The signal processing operations of the CELP decoder are:
following operations occur: 1. Select the appropriate code word for the current frame from
The codebook vector is scaled by the LPC gain estimate, a matching excitation VQ codebook
yielding the error signal, e[n] 2. Scale the code word sequence by the gain of the frame,
The error signal, e[n], is used to excite the long-term pitch thereby generating the excitation signal, e[n] process e[n] by
predictor, yielding the estimate of the speech signal, x[n], for the long-term synthesis filter and the short-term vocal tract
the current codebook vector filter, giving the estimated speech signal x[n].
3. Process the estimated speech signal by an adaptive post filter
The signal, d[n], is generated as the difference between the
whose function is to enhance the formant regions of the speech
speech signal, x[n], and the estimated speech signal, x[n]
signal, and thus to improve the overall quality of the synthetic
The difference signal is perceptually weighted and the speech from the CELP system
resulting mean-squared error is calculated
 Features, Advantages and Disadvantages
Features
1. Linear predictive coding of speech is a very * Selective speech computation.
powerful speech analysis technique used to estimate * Good quality speech at low bit rate.
the basic speech parameters such as
Pitch(Frequency), Formants(Enhanced frequency), Advantages
Vocal Tract(Mouth and Throat) and  The primary advantage of Code Excited Linear Predictive coding
Intensity(Loudness). (CELP) is that it reduces the bitrates of the speech i.e. the size of
2. Linear predictive coding provides: the transmitting signal is reduced.
* Extremely accurate estimation of speech  Provides significantly better quality than existing low bit-rate
parameters. algorithms, such as Residual-Excited Linear Prediction (RELP) and
Linear Predictive Coding (LPC)
 The signal transmitted through CELP requires less bandwidth and
hence number of users can be increased.
 CELP coding uses the encryption of data so the data is secured until
Disadvantages the destination.
 Due to reduce in the bitrates of the speech signal,  The CELP speech sounds much closer to the original one (still with

the quality of voice signal is reduced. a muffled sound) as compared with the LPC
 This technique is loss compression technique,  For each segment, encoder finds excitation vector that generates

hence data gets faded if transmitted to the long synthesized speech that best matches speech being coded.
distance.
 The CELP method has inherent errors and in most
cases doesn’t give accurate solutions.
Importance in today’s world and some Variations
1. In the todays world CELP is most widely used for speech coding algorithm.
2. The quality of a speech attainable using CELP and the ease of a real time implementation with single-chip
DSPs has led to widespread implementations in communication and voice storage system.
3. In real-time implementation on a fixed-point DSP is desirable because of its lower cost and power
consumption compared with floating-point DSPs. However, the limited dynamic range of the fixed point
processor leads to a loss in precision and hence a loss in performance.
4. It is also mainly used to construct vocoders where musical instruments are used as excitation signal to the time-
varying filter estimated from a singer's speech. This is somewhat popular in electronic music.
5. Below are some variations of Code Excited Linear Prediction speech coding algorithm:
* Low Delay (LD) CELP Coder.
* ACELP : Algebraic Code Excited Linear Prediction.
* CS- ACELP : Conjugate Structure ACELP.
* VSELP : Vector-Sum Excited Linear Predictive Coding.
* EVSELP : Enhanced VSELP.
* PSI-CELP : Pitch Synchronous Innovation-Code Excited Linear Prediction.
* RPE-LTP : Regular Pulse Exciting-Long Term Prediction-linear predictive coder.
* MP-MLQ : Multipulse-Maximum Likelihood Quantization.
Applications

1. Network-64 Kbps PCM (8 kHz sampling rate, 8- bit log quantization).

2. International-32 Kbps ADPCM.
3. Teleconferencing-16 Kbps LD-CELP.
4. Wireless-13, 8, 6.7, 4 Kbps CELP-based coders.
5. Secure telephony-4.8, 2.4 Kbps LPC-based coders (MELP).
6. VoIP-8 Kbps CELP-based coder.
7. Storage for voice mail, answering machines, announcements-16 Kbps LC-CELP.
 Conclusion

1. The basic principle of Code Excited Linear Prediction algorithm for speech coding is that the speech samples can be
approximated as linear combination of the past samples by minimizing the error. This can be done by minimizing the
sum of squared differences between the actual speech samples and linearly predicted ones.
2. A coder compresses the source data by removing redundant information. The receiver must know the compression
mechanism to decompress the data.
3. The graphical results of ADPCM (Adaptive delta pulse code modulation) decoder and decoding using CELP is shown
below:

using ADPCM using CELP

It can be observed from comparative study of waveforms of both coders that CELP reproduces the signal more closely to
the original signal as compared to other coders.
4. Hence we can conclude that CELP is superior over other waveform coders in terms of signal reproduction and also in
terms of bit rate and reduction in bandwidth requirement.
 References
* J.L. Flanagan and L. R. Rabiner Speech Synthesis, Dowden, Hutchington & Ross, Inc., Stroudsburg, Pennsylvania 1973.
* IJESRT - LINEAR PREDICTIVE CODING Mr. Mohit Narayanbhai Raja, Miss. Priyanka Richhpal Jangid, Dr. Sanjay M. Gulhane.
* https://www.youtube.com/watch?v=7qbxqFlUAlI
* http://www.icg.isy.liu.se/courses/tsbk01/material/slides/11_SpeechCoding.pps
* https://www.slideshare.net/sdd2311/lpc-29316427
* http://www.ijrdet.com/files/Volume3Issue1/IJRDET_0714_28.pdf
* https://en.wikipedia.org/wiki/Code-excited_linear_prediction#:~:text=Code%2Dexcited%20linear%20prediction%20
* https://www.researchgate.net/publication/273452039_Speech_Coding_Using_Code_Excited_Linear_Prediction

-------------X-------------