Scribd
Upload
106 views

Filter Implementation Using DSP Kit

This document outlines the goals of understanding speech generation in humans, exploring speech enhancement techniques, and implementing a low pass filter on an FPGA kit. It discusses various speech enhancement methods including statistical model-based algorithms, spectral subtraction algorithms, and subspace algorithms. Spectral subtraction and multi-band spectral subtraction methods are explained in more detail. An introduction to the Spartan III DSP kit is provided along with design steps for generating VHDL code, loading a bit file, and testing a low pass filter on the kit by filtering noisy speech in real time.

Uploaded by

Pravin Gondaliya
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
106 views

Filter Implementation Using DSP Kit

This document outlines the goals of understanding speech generation in humans, exploring speech enhancement techniques, and implementing a low pass filter on an FPGA kit. It discusses various speech enhancement methods including statistical model-based algorithms, spectral subtraction algorithms, and subspace algorithms. Spectral subtraction and multi-band spectral subtraction methods are explained in more detail. An introduction to the Spartan III DSP kit is provided along with design steps for generating VHDL code, loading a bit file, and testing a low pass filter on the kit by filtering noisy speech in real time.

Uploaded by

Pravin Gondaliya
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 16

Guided by: Pro.

Tanish Zaveri Pravin Gondaliya(08BEC029) Surendra Jalu(08BEC029)

Goal
To understand the model of speech generation in

human beings. To understand speech enhancement techniques. To explore the features and peripherals of the Spartan III DSP kit. Implement a low pass filter using VHDL coding. Filter the speech signal on FPGA kit in real time.

Scopes of Acoustic Science

Speech Enhancement
Speech enhancement is concerned with improving some perceptual aspect of speech that has been degraded by additive noise. Different kind of noise affect on the quality of the speech. Different speech enhancement techniques are used to improve the quality of speech and reduce the specific noise coming from different sources at different SNRs.

Goals and Classes of Speech Enhancement


Speech enhancement has three major goals: To improve the quality and intelligibility of speech corrupted by background noise. To make speech coders robust when to input noise. To make speech recognition systems more robust to input noise.

Cont
Speech enhancement techniques are divided in main three classes. Statistical model based algorithms Wiener a priori SNR method Wavelet thresholding method Log MMSE estimator method Spectral subtractive algorithms Spectral subtraction algorithm Multiband spectral subtractive algorithm Subspace algorithms

Wavelet Thresholding Method


Wavelet-based speech enhancement algorithm can be summarized by the following three steps : Wavelet transform of noisy signal Thresholding the resulting wavelet coefficients Inverse transform to obtain the denoised signal Hard thresholding , Soft thresholding, Non linear thresholding

Spectral Subtraction Method


It is one of the most popular methods of reducing the effect of background (additive) noise. Assume that y(n), noise corrupted input signal is composed of the clean speech signal x(n) and additive noise signal d(n) : y(n) = x(n) + d(n) Taking Discrete Fourier Transform on both sides Y () = X() + D() Y() in polar form : Y() = |Y()| exp(jy ()) |Y()| is the magnitude and exp(jy ()) is the phase spectrum of the corrupted noisy signal.

Block Diagram

Multi Band Spectral Subtraction(MBSS) Method


Noise signal does not affect the speech signal uniformly

over the whole spectrum. Some frequencies are affected more adversely than the others depending on the spectral characteristics of the noise.. So in MBSS Speech is processed into N (1 N 8) overlapping frequency bands and spectral subtraction is performed independently on each band using bandspecific over-subtraction factors . This method provides a greater degree of flexibility and control on the noise subtraction levels that reduces artifacts in the enhanced speech, resulting in improved speech quality.

Introduction of kit
The Spartan 3A DSP reference kit is an integrated circuit development platform for Xilinxs Spartan 3A DSP FPGA.
189Kbytes of management.

block

RAM,

Digital

clock

The reference board also features external non volatile memory (SPI flash) and volatile (DDR SDRAM) memory, power supplies, LCD ports, Audio, Camera, Video Interfaces and other I/O devices.

Cont
FPGA Configuration Options
1) SPI Serial Flash 2) JTAG

Buttons and Switches


8 position DIP switch 8 Discrete LEDs One push button Switch Push Button Switch to initiate FPGA programming.

Design steps
Generate VHDL code using matlab FDA tool.
Create a project in ISE project navigator and add

VHDL file to project. Synthesize and implement the file. Generate bit file for loading into the kit. Load bit file using iMPACT. Test the results.

Design
Give speech input to one of the stereo channel and

noise tone to other channel. These two signals will be added together and generate a noisy speech. This speech will be filtered by suitable low pass filter and given to the speaker.

THANK YOU

You might also like