Acoustic Echo Cancellation using LMS Algorithm

Transcription

1 Acoustic Echo Cancellation using LMS Algorithm Nitika Gulbadhar M.Tech Student, Deptt. of Electronics Technology, GNDU, Amritsar Shalini Bahel Professor, Deptt. of Electronics Technology,GNDU,Amritsar Harmeet Kaur Assistant Professor,Deptt. of Electronics Technology, GNDU, Amritsar ABSTRACT One of the core technologies these days in digital signal processing which find numerous application areas in science as well as in industry is adaptive filtering. Adaptive filtering techniques are used in a wide range of applications, including adaptive equalization, echo cancellation, adaptive noise cancellation, and adaptive beam forming. In today s telecommunication system, acoustic echo cancellation is a common occurrence. The signal interference caused by acoustic echo is distracting the users and causes a reduction in the quality of the voice communication. Echo cancellers are very successful these days and almost no echo can be perceived while using telephones. The present work focuses on the LMS algorithm to reduce the unwanted echo which results in improving communication quality. LMS algorithm is a type of adaptive filter algorithm and is used to determine the minimum mean square error and is based on steepest descent method and gradient search technique. Thus in this research work we are going to eliminate noise signal from audio signal using this LMS algorithm. This causes the clear and high quality audio signal as an output signal. Simulation results are presented to support the analysis. They show that the LMS algorithm reduced the echo from the input noisy signal by 40 percent. KEYWORDS Adaptive filtering, Echo, Acoustic Echo Cancellation, LMS Algorithm I. INTRODUCTION The basic need of the modern world is effective voice and audio communication. In acoustic applications noise reflected from surrounding environment reduces quality of audio signal and sometimes it becomes impossible to recover the original speech or audio signal which was transmitted. To remove noise from the noise contained signal and to enhance the quality of signal, the technique which gain much attention is Acoustic Noise Cancellation (ANC) [1]. An adaptive filter is an important block of ANC and is a filter that self-adjusts its transfer function according to its optimization algorithm which is driven by an error signal. It had been firstly proposed by Kelly of Bell Telephone Laboratories around 1965 [2]. It adjusts its filter coefficients automatically to adapt the input signal via an adaptive algorithm. Adaptive filters are used in situation where speech and noise signals are random in nature. Fig 1. Block diagram of Adaptive filter Adaptive filters play an important role in modern digital signal processing (DSP) products in areas such as telecommunication for echoes in the transmission channel, for the cancellation of noise and also in digital controller for active noise control. It is also used in the field of biomedical, image signal processing and 140 Nitika Gulbadhar, Shalini Bahel, Harmeet Kaur

2 sonar. Thus with the proper algorithm adaptive filter can operate with different conditions and readjusts itself continuously to minimise the error signal. Bernard widrow et al. first introduced the basic concept of adaptive filters [3]. Figure 1 shows the basic block diagram of adaptive filter. Functioning of various blocks are as follows: Filter Structure This is the implementation of the filtering algorithm. This block computes the filter output with respect to the input. The filter coefficients can be updated by the Adaptive Algorithm. Criterion of Performance If we know the desired response then this block compare it to the actual response and then indicate to the Adaptive Algorithm that something needs to be changed. Adaptation Algorithm The function of this block is to adjust the filter coefficients according to the output of criterion of performance. Echo is a reflected version of original waveform being reflected and attenuated [4]. In telecommunication, echo can severely affect the quality and intelligibility of voice conversation in a telephone system. In general, the noticeable echoes have an appreciable amplitude and a delay of more than 1 ms. Echo will give telephone call a sense of liveliness provided the round-trip delay is of the order of a few milliseconds. However, with the increasing amplitude and also round trip delay of more than 20 ms echoes become increasingly annoying and objectionable. Hence an important aspect of the design of modern telecommunication systems is echo cancellation [5]. Although there are other ways of cancelling echo such as echo cancellers, echo barriers, echo suppressors but adaptive filters are widely used due to its stability and wide scope of improvements. II. ACOUSTIC ECHO CANCELLATION An adaptive filter minimizes the function of difference between the desired output d(n) & actual output y(n),by altering its parameters regularly. The block diagram of the adaptive echo cancellation system is shown in figure 2. Here, w(n) represents the adaptive filter used to cancel the echo signal and h(n) represents the impulse response of the acoustic environment which generates the echo. The adaptive filter tries to equate its output y(n) to the desired output d(n) (the echo signal generated within the acoustic environment). After every iteration, the error signal i.e. e(n) = d(n) y(n) is generated which is given as a feedback to the filter. Adaptive algorithm guides the change in filter parameters. When the adaptive filter output is equal to desired signal, the error signal ideally becomes zero and far user will not experience any kind of disturbance [6]. Error Signal e(n) = d(n) y(n) Fig 2. Adaptive Echo Cancellation System III. LMS ALGORITHM LMS is initially proposed by widrow Hoff in It is used to determine the minimum mean square error and is based on steepest descent method and gradient search technique. Mostly this algorithm is used because of ease of implementation, simplicity and low computational complexity. If x(n) is the input signal vector and w(n) is the weight vector of the adaptive filter then, output of adaptive filter y(n) is given by y(n) = w(n) T x(n) (1) error signal e(n) is given by e(n) = d(n) y(n) (2) 141 Nitika Gulbadhar, Shalini Bahel, Harmeet Kaur

3 weight vector update equation is given by w(n+1) = w(n) + µe(n)x(n) (3) where µ is the step size and controls the convergence rate. Small value of µ leads to more convergence time. Large value of µ cause the algorithm to diverge and degrades the performance of adaptive filter. Therefore selecting a step size is very important. One of the primary disadvantages of LMS algorithm is a fixed step size for every iteration. This requires an prior understanding of statistics of input signal which is rarely achievable. Also in LMS algorithm the correction that applied to w(n) is proportional to the input vector x(n). Therefore when x(n) is large, the LMS algorithm experiences a problem with noise gradient amplification [7]. IV. SIMULATION AND RESULTS This section involves the simulation of LMS for acoustic echo cancellation. In this simulation we take a signal and corrupted by noise called adaptive white gaussian noise. Then LMS adaptive filtering is used to estimate and remove the echo from the distorted signal, thus, creating a reconstructed signal. This algorithm seek to minimize the excess mean-square error (MSE) between the echo signal and the estimated echo. Simulations of the LMS algorithm will be done in MATLAB. Different parameters such as filter output, error signal, input signal with added noise, MSE are plotted. Fig 3. Input and Input noisy signal Figure 3 shows the input signal with added noise and input signal without noise in sinusoidal form. 142 Nitika Gulbadhar, Shalini Bahel, Harmeet Kaur

4 Fig 4. Output signal and Error plot Figure 4 shows the received output after applying LMS filtering and the error removed by the filter i.e, the error computed. Fig 5. Signal plot, Output plot of echo cancellation, MSE plot 143 Nitika Gulbadhar, Shalini Bahel, Harmeet Kaur

5 Figure 5 shows the echo cancellation by applying LMS algorithm. Here learning curve shows the MSE error which decreases by increasing the number of iterations. A learning curve is shows the MSE approaching 40 db after approximately one hundred iterations. Figure 5 also shows the input signal added to the echo signal and plotted under it is the reconstructed signal. After approximately a hundred iterations, the original input signal is restored from the distorted signal. V. CONCLUSION Thus we have implemented the new type of filtering scheme for noise cancellation from audio signal. LMS filter has increased efficiency and reduced time consumption for filtering. Echo is removed from the signal by using the proposed LMS filter by measuring the basic parameters like frequency and rate of sample of the audio signal. Thus the proposed system made the audio signal as noise free. Matlab provided the software module to perform the algorithm and Graphical User Interface showed the results. This algorithm reduced the echo from the original signal by 30 percent. This algorithm can be used in practical applications like inverse modeling, adaptive feedback cancellation in hearing aids and jammer suppression etc. REFERENCES [1] Ms. Mugdha. M. Dewasthale and Dr. R. D. Kharadkar, Acoustic Noise Cancellation using Adaptive Filters: A Survey, in proceedings of 2014 International Conference on Electronic Systems, Signal Processing and Computing Technologies, pp (DOI: ). [2] Michael Reuter, Nonlinear Effects in LMS Adaptive Equalizers, IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 47, NO. 6, JUNE [3] Bernard widrow, John Glover, Adaptive Noise Cancelling : Principles and Applications, IEEE proceeding, Vol. 63 No 12,1975. [4] Upal Mahbub, Shaikh Anowarul Fattah, A Single-Channel Acoustic Echo Cancellation Scheme Using Gradient- Based Adaptive Filtering, in proceedings of Circuits Syst Signal Process(2014) (DOI: ). [5] Vladimir M. Matic and Srdan N. Abadzic, "Acoustic and line echo cancellation using adaptive filters", 15th Telecommunications forum TELFOR. pp , November [6] Abhishek Deb, Asutosh Kar and Mahesh Chandra, A Technical Review on Adaptive Algorithms for Acoustic Echo Cancellation, in proceedings of International Conference on Communication and Signal Processing, April 3-5, 2014, India, pp [7] Nitika Gulbadhar, Shalini Bahel, Harmeet Kaur, Adaptive Algorithms for Acoustic Echo Cancellation : A Review, in proceedings of International Journal of Engineering Trends and Technology (IJETT) Volume 34 Number 5- April 2016,pp Nitika Gulbadhar, Shalini Bahel, Harmeet Kaur