Scribd Logo
Upload

Welcome to Scribd!

  • 67 views

    Acoustic Echo Cancellation

    Uploaded by

    Subhadeep Dash
    This document discusses acoustic echo cancellation using adaptive filters. It introduces acoustic echo and how adaptive filters can be used to cancel echo. The LMS and NLMS algorithms are described as methods to update the adaptive filter weights to minimize the error between the desired and actual signals. The document outlines the implementation of the LMS and NLMS algorithms for echo cancellation, and presents results showing the NLMS algorithm exhibits better convergence properties than LMS while having only slightly higher computational complexity.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    Acoustic Echo Cancellation

    Uploaded by

    Subhadeep Dash
    67 views10 pages
    This document discusses acoustic echo cancellation using adaptive filters. It introduces acoustic echo and how adaptive filters can be used to cancel echo. The LMS and NLMS algorithms are described as methods to update the adaptive filter weights to minimize the error between the desired and actual signals. The document outlines the implementation of the LMS and NLMS algorithms for echo cancellation, and presents results showing the NLMS algorithm exhibits better convergence properties than LMS while having only slightly higher computational complexity.

    Original Description:

    Acoustic echo cancellation using adaptive filters

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses acoustic echo cancellation using adaptive filters. It introduces acoustic echo and how adaptive filters can be used to cancel echo. The LMS and NLMS algorithms are described as methods to update the adaptive filter weights to minimize the error between the desired and actual signals. The document outlines the implementation of the LMS and NLMS algorithms for echo cancellation, and presents results showing the NLMS algorithm exhibits better convergence properties than LMS while having only slightly higher computational complexity.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    67 views10 pages

    Acoustic Echo Cancellation

    Uploaded by

    Subhadeep Dash
    This document discusses acoustic echo cancellation using adaptive filters. It introduces acoustic echo and how adaptive filters can be used to cancel echo. The LMS and NLMS algorithms are described as methods to update the adaptive filter weights to minimize the error between the desired and actual signals. The document outlines the implementation of the LMS and NLMS algorithms for echo cancellation, and presents results showing the NLMS algorithm exhibits better convergence properties than LMS while having only slightly higher computational complexity.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 10

    ACOUSTIC ECHO

    CANCELLATION

    DIGITAL SIGNAL ANALYSIS AND APPLICATIONS PROJECT - GROUP 13


    GROUP MEMBERS:
    1. HINDUJA BIKKI - 201601013
    2. DASH SUBHADEEP – 201601021
    3. SUBASH KARTHIKEYA - 201601052
    4. SAI VAMSHI - 201601080
    INTRODUCTION

    ACOUSTIC ECHO OCCURS WHEN AN


    AUDIO SIGNAL IS REVERBERATED IN A REAL
    ENVIRONMENT. ADAPTIVE FILTERS ARE
    DYNAMIC FILTERS WHICH ITERATIVELY ALTER
    THEIR CHARACTERISTICS IN ORDER TO
    ACHIEVE AN OPTIMAL DESIRED OUTPUT. THE
    AIM OF AN ADAPTIVE FILTER IS TO CALCULATE
    THE DIFFERENCE BETWEEN THE DESIRED
    SIGNAL AND THE ADAPTIVE FILTER OUTPUT.
    IN CASE OF ACOUSTIC ECHO
    CANCELLATION, THE OPTIMAL OUTPUT OF
    THE ADAPTIVE FILTER IS EQUAL IN VALUE TO
    THE UNWANTED ECHOED SIGNAL. IF THE
    ADAPTIVE FILTER OUTPUT IS EQUAL TO
    DESIRED SIGNAL, THE ERROR SIGNAL GOES
    TO ZERO. IN THIS SITUATION, THE ECHOED
    SIGNAL WOULD BE COMPLETELY CANCELLED
    AND THE FAR USER WOULD NOT HEAR ANY
    OF THEIR ECHOED SPEECH RETURNED TO
    THEM
    WORKING

     THE INPUT SIGNAL IS SENT BY THE USER TO THE PROJECT(REALTIME


    RECORDING).
     ECHO IS ADDED TO THE INPUT SIGNAL THROUGH MATLAB AND THIS
    OBTAINED SIGNAL IS SENT FOR FURTHER PROCESSING.
     THEN BY USING SUITABLE ADAPTIVE FILTERS, WE ARE ABLE TO REMOVE THE
    UNWANTED ECHO SIGNAL.
     LMS AND NLMS ALGORITHMS ARE BEING USED HERE TO OBTAIN THE DESIRED
    ECHO CANCELLED SIGNAL.
     AFTER PROCESSING, THE USER CAN PLAY THE ECHO CANCELLED SIGNAL
    AND VERIFY THE RESULTS.
    LEAST MEAN SQUARE ALGORITHM(LMS)

     LMS ALGORITHM IS WELL KNOWN AND WIDELY USED DUE TO ITS


    COMPUTATIONAL SIMPLICITY.
     IT IS THIS SIMPLICITY THAT HAS MADE IT THE BENCHMARK AGAINST WHICH ALL
    OTHER ADAPTIVE ALGORITHMS ARE JUDGED.
     WITH EACH ITERATION OF THE LMS ALGORITHM, THE FILTER TAP WEIGHTS OF
    THE ADAPTIVE FILTER ARE UPDATED.
     THE STEP SIZE PARAMETER(𝜇) IN THIS ALGORITHM, WHICH IS A SMALL POSITIVE
    CONSTANT, CONTROLS THE INFLUENCE OF THE UPDATING FACTOR.
     IF THE VALUE OF THE STEP SIZE PARAMETER (𝜇) IS TOO LARGE, THE ADAPTIVE
    FILTER BECOMES UNSTABLE AND THE OUTPUT DIVERGES.
    NORMALIZED MEAN SQUARE ALGORITHM
    (NLMS)

     ONE OF THE PRIMARY DISADVANTAGES OF THE LMS ALGORITHM IS HAVING


    A FIXED STEP SIZE.
     THE NORMALISED LEAST MEAN SQUARE ALGORITHM(NLMS) IS AN EXTENSION
    OF THE LMS ALGORITHM WHICH BYPASSES THIS ISSUE BY CALCULATING
    MAXIMUM STEP SIZE VALUE.
     THE STEP SIZE IS THE RECIPROCAL OF TWICE THE DOT PRODUCT OF THE INPUT
    VECTORS.
     AS THE STEP SIZE PARAMETER IS CHOSEN BASED ON INPUT VALUES, THE NLMS
    ALGORITHM TENDS TO SHOW FAR GREATER STABILITY WITH UNKNOWN
    SIGNALS.
     THE GOOD CONVERGENCE SPEED AND RELATIVE COMPUTATIONAL
    SIMPLICITY MAKE NLMS IDEAL FOR ECHO CANCELLATION SYSTEM.
    IMPLEMENTATION OF LMS ALGORITHM

     THE IMPLEMENTATION OF LMS ALGORITHM INVOLVES THREE STAGES FOR


    EACH ITERATION.
     THE OUTPUT OF THE ADAPTIVE FILTER IS OBTAINED BY THE DOT PRODUCT OF
    THE FILTER TAP WEIGHTS WITH THE ECHOED SIGNAL.
    𝑦 𝑛 = 𝑤 𝑇 𝑛 . 𝑥(𝑛)
     THE VALUE OF THE ERROR IS CALCULATED AS FOLLOWS.
    𝑒 𝑛 = 𝑑 𝑛 − 𝑦(𝑛)
     THE TAP WEIGHTS OF THE FIR FILTER ARE UPDATED IN PREPARATION OF THE
    NEXT ALGORITHM.
    𝑤 𝑛 + 1 = 𝑤 𝑛 + 2𝜇𝑒 𝑛 𝑥(𝑛)
    IMPLEMENTATION OF NLMS ALGORITHM

     THE IMPLEMENTATION OF NLMS ALGORITHM INVOLVES FOUR STAGES FOR


    EACH ITERATION.
     THE OUTPUT OF THE ADAPTIVE FILTER IS OBTAINED BY THE DOT PRODUCT OF
    THE FILTER TAP WEIGHTS WITH THE ECHOED SIGNAL.
    𝑦 𝑛 = 𝑤 𝑇 𝑛 . 𝑥(𝑛)
     THE VALUE OF THE ERROR IS CALCULATED AS FOLLOWS.
    𝑒 𝑛 = 𝑑 𝑛 − 𝑦(𝑛)
     THE STEP SIZE FOR THE INPUT VECTOR IS CALCULATED.
    𝜇 𝑛 = 1/𝑥 𝑇 𝑛 𝑥(𝑛)
     THE TAP WEIGHTS OF THE FIR FILTER ARE UPDATED IN PREPARATION OF THE
    NEXT ALGORITHM.
    𝑤 𝑛 + 1 = 𝑤 𝑛 + 𝜇𝑒 𝑛 𝑥(𝑛)
    RESULTS
    CONCLUSION

     THE LMS ALGORITHM HAS 2N + 1 MULTIPLICATIONS WHILE NLMS HAS 3N+1


    MULTIPLICATIONS.
     N FOR OBTAINING THE OUTPUT SIGNAL, ONE FOR SCALAR MULTIPLICATION
    OF 2𝜇𝑒 𝑛 AND N FOR THE SCALAR BY VECTOR MULTIPLICATION AND AN
    EXTRA N IN CASE OF NLMS FOR CALCULATION OF 𝜇.
     BECAUSE OF ITS SIMPLICITY, THE LMS ALGORITHM IS THE MOST POPULAR
    ADAPTIVE ALGORITHM.
     HOWEVER IT SUFFERS FROM SLOW AND DATA DEPENDENT CONVERGENCE
    BEHAVIOUR.
     NLMS BEING EQUALLY SIMPLE, EXHIBITS A BALANCE BETWEEN SIMPLICITY AND
    PERFORMANCE THAN THE LMS ALGORITHM.
    THANKS FOR READING

    REFERENCE:
    https://pdfs.semanticscholar.org/14dc/1a8b346c41cd01279e54aeeccb9887fa
    174e.pdf

    You might also like