Echo Cancellation Final Ppt New.pptx

ACOUSTIC ECHO CANCELLATION USING OPTIMIZED NLMS ALGORITHM R.Madhubala (812012106045) S.Pradeepa (8121012106054) C.Priyadharshini (812012106056) under the guidance of Mr.S.Syed Shaffi Associate Professor Department of Electronics and Communication Engineering M.A.M. College of Engineering and Technology.

OUTLINE Abstract Introduction Literature Review Objective Block diagram Functional diagram of adaptive filter Simulation outputs Advantages Applications Conclusion References 12/4/2016

MAMCET

2

ABSTRACT

ABSTRACT In

acoustic

echo

environment,

the

quality

of

communication is reduced. To improve the quality , many adaptive techniques

are used for echo cancellation. Some of them are: 1) LMS algorithm 2) RLS algorithm 3) Kalman

filter

4) NLMS algorithm

In this system, Optimized NLMS algorithm is used. 12/4/2016

MAMCET

4

INTRODUCTION

SIGNAL PROCESSING AND ACOUSTIC ENVIRONMENT Signal Processing is the intentional alteration of audio signals in order to achieve the desired output. Acoustic echo is an environment in which sounds from loudspeaker being reflected and recorded by microphone which varies with time. In this, the original signal is affected by unwanted noise and echo signals. The actual output is different from that of desired output in the acoustic environment. 12/4/2016

MAMCET

6

ORIGIN OF ACOUSTIC ENVIRONMENT

12/4/2016

MAMCET

7

OPTIMIZED NLMS ALGORITHM The optimized NLMS algorithms follows jointoptimization problem on both normalized step-size

and regularization parameters. The convergence rate always decreases when the regularization constant increases. The misadjustment increases when regularization constant decreases. It does not require any additional parameters to control. 12/4/2016

MAMCET

8

LITERATURE REVIEW

LEAST MEAN SQUARE ADAPTIVE FILTER FOR ACOUSTIC ECHO CANCELLATION TECHNIQUE USED:

LMS algorithm

In this paper, it utilises the LMS adaptive filter to perform echo cancellation by modifying gradient vector of the

filter tap weights to obtain optimal output signal. DRAWBACKS: Fixed step size Sensitive to the changes in the input S.Haykin, B. Widrow (Eds.), 2003 12/4/2016

MAMCET

10

ON REGULARIZATION IN ADAPTIVE FILTERING TECHNIQUE USED: NLMS, SR-NLMS, IP-NLMS, SR-

IPNLMS. In this paper, four adaptive algorithms are proposed for the derivation of an optimal regularization parameter and these adaptive algorithms behave extremely well at all ENR levels DRAWBACKS: More misalignment and never converge. Depends on regularization parameter and ENR values. J.Benesty, C.Paleologu, S.Ciochina, 2011 12/4/2016

MAMCET

11

VARIABLE STEP SIZE NLMS AND AFFINE PROJECTION ALGORITHMS TECHNIQUE USED: VSS-NLMS and AP algorithm In this paper, step size of AP algorithm is predetermined

by considering the cross correlation between the current weight error vector and prior measurement noises with input signal & VSSNLMS is used for faster convergence.

DRAWBACKS: Selection of step size is difficult Misadjustment becomes larger H.C. Shin, A.H. Sayed, W.-J. Song , 2012 12/4/2016

MAMCET

12

STUDY OF GENERAL KALMAN FILTER FOR ECHO CANCELLATION TECHNIQUE USED:

Kalman Filter

In this paper, it utilizes Kalman filter for acoustic echo cancellation by considering block of time samples for each iteration and also it shows about its connection to adaptive filters. DRAWBACKS: Inaccurate initialization

Computationally expensive C.Paleologu, J.Benesty, S.Ciochina, 2013 12/4/2016

MAMCET

13

OBJECTIVE To

improve

the quality of

communication by

removing the unwanted signals such as echo, noise and interference signals by using optimized NLMS

adaptive algorithm technique and hence to get the optimized output.

12/4/2016

MAMCET

14

BLOCK DIAGRAM FOR ACOUSTIC ECHO CANCELLATION Input signal x(n)

Adaptive filter w(n)

Acoustic Impulse Response h(n)

+

Error signal e(n)= d(n)-y(n) 12/4/2016

Echoed signal d(n) MAMCET

15

FUNCTIONAL DIAGRAM OF ADAPTIVE FILTER

12/4/2016

MAMCET

16

SOFTWARE USED

MATLAB version 8.1

12/4/2016

MAMCET

17

SIMULATION RESULTS Impulse response Original speech signal Echo signal Interference signal Noise signal

Microphone signal Filter output Updating of weighted coefficients

Error reduction 12/4/2016

MAMCET

18

IMPULSE RESPONSE

12/4/2016

MAMCET

19

ORIGINAL SPEECH SIGNAL

12/4/2016

MAMCET

20

ECHOED SPEECH SIGNAL

12/4/2016

MAMCET

21

NOISE SIGNAL

12/4/2016

MAMCET

22

INTERFERENCE SIGNAL

12/4/2016

MAMCET

23

MICROPHONE SIGNAL

12/4/2016

MAMCET

24

FILTER OUTPUT

12/4/2016

MAMCET

25

UPDATING OF WEIGHTED COEFFICIENTS

12/4/2016

MAMCET

26

REDUCTION OF ERROR

12/4/2016

MAMCET

27

COMPARISON OF VARIOUS ALGORITHMS Algorithm

LMS

NLMS

RLS

12/4/2016

Multiplicatio n operations

Advantages & disadvantages

2N+1

Simple requires prior knowledge of the input

3N+1

computationally efficient stable performance with nonstationary signals.

(4N)^2

Good attenuation Computational complexity

MAMCET

28

ADVANTAGES Faster convergence and tracking Low misadjustment Improved system performance Non-parametric in nature More stability

12/4/2016

MAMCET

29

APPLICATIONS

Hands-free communications: Mobile telephony Teleconferencing systems

12/4/2016

MAMCET

30

CONCLUSION An optimized NLMS algorithm is proposed. Following a convergence analysis of the NLMS adaptive filter, the JO-NLMS algorithm is based on

joint-optimization on both the normalized step-size and regularization parameters. The

proposed

algorithm

achieves

faster

convergence rate and provides almost perfect results for cancelling residual echoes.

12/4/2016

MAMCET

31

FUTURE ENHANCEMENT In real life situations, multichannel sound is the

norm for telecommunication. In order to handle such situations in a better way the echo cancellation algorithm developed should be

extended for the multichannel case.

12/4/2016

MAMCET

32

REFERENCES Benesty.J,

Paleologu.C, Ciochina.S (2011) , On

regularization

in

adaptive

filtering,

IEEETrans.AudioSpeechLang.Process. Chang.M, Kong.N (2009) ,Scheduled-stepsize NLMS algorithm, IEEE Signal Process.Lett.16.

Enzer.G, Vary.P (2006) Frequency- Domain adaptive Kalman filter for acoustic echo control in hands-free telephones, Signal Process.

Haykin.S,

Widrow.B(Eds), Wiley, Hoboken (2003) ,

Least-Mean –Square Adaptive Filters. 12/4/2016

MAMCET

33

THANK YOU 12/4/2016

MAMCET

34