International Journal of Research in Advent Technology, Vol.2, No.6, June 2014 E-ISSN: 2321-9637
Acoustic Echo Cancellation from the Signal Using LMS Algorithm Ashu Sharma1, Yogesh Juneja2 Electronics and communication1, 2, PDM college of Engg1, 2 Email: kaushik.ashu12@gmail.com1, yogeshjunejaer@gmail.com2
Abstract- The term Acoustic Echo Cancellation (AEC) refers to a process of removing echo from the received signal that contains one or more delayed signals (copies of the original signal). The primary step while cancelling an echo is to identify the transmitted signal which reappears with some delay. Once the echo is identified it is cancelled by subtracting from transmitted signal. Echo cancellation can be done using either echo suppressors or echo cancellers, or in some case both. But suppressors support only half duplex communication leading to the invention of echo cancellers which allows both the speakers to talk at the same time. This paper is concerned with Adaptive Acoustic Echo Cancellation Based on Least Mean Square (LMS) Algorithm. Here, we evaluate the performance of telecommunication systems like hands-free and teleconferencing systems which is affected by white noise. Index Terms- LMS, NLMS, AEC 1. INTRODUCTION Adaptive filters are a type of digital filters that have self optimizing characteristics. Such filters have a finite number of parameters that are adjusted by adaptive algorithms to optimize some performance criteria. From last few decades adaptive filtering is gaining momentum in many Digital signal processing (DSP) applications. Digital signal processing (DSP) has been a major player in the current technical advancements such as noise filtering, system identification, and voice prediction. Standard DSP techniques, however, are not enough to solve these problems quickly and give acceptable results. Adaptive filtering techniques must be implemented for accurate solutions. An adaptive filter is a computational device that attempts to model the relationship between two signals in real time in an iterative manner. Adaptive filters are often realized either as a set of program instructions running on an arithmetical processing device such as a microprocessor or DSP chips[1]. An adaptive filter is defined by four aspects: 1. Signals being processed by the filters. 2. The structure that defines how the output signal of the filter is computed from its input signal. 3. The parameters within this structure that can be iteratively changed to alter the filter’s input-output relationship. 4. The adaptive algorithm that describes how the parameters are adjusted from one time instant to the next. Adaptive Filter The block diagram of an adaptive filter is as shown in fig. 1 [2]. It is the adaptive algorithm that utilizes the
coefficient updation according to the coefficient update equation of the form
…(1) Where ∆Wn is a correction that is applied toWn at time n to form a new Wn+1 at time (n+1). The keycomponent of adaptive algorithm is, how the correction ∆Wn to be formed [3].
Fig 1 Adaptive Filter
2. SYSTEM DESCRIPTION Acoustic echo cancellation using adaptive algorithms Acoustic echo control is widely used application area. Designing adaptive filters is an important task, Adaptive filters are dynamic filters which iteratively alter their characteristics in order to achieve an optimal desired output[4]. An adaptive filter algorithmically alters its parameters in order to minimize a function of the difference between the desired output d(n) and its actual output y(n). This function is known as the cost function of the adaptive algorithm [5].
144
International Journal of Research in Advent Technology, Vol.2, No.6, June 2014 E-ISSN: 2321-9637 The aim is to cancel the desired input signal d(n) by making sure the error signal e(n) is kept to the best minimum value possible. From fig. 2 it is noted that the past values of the estimation error signal e(n) is fed back to the adaptive filter. The purpose of the feedback is to effectively adjust the structure of the adaptive system, thus altering its response characteristics to the optimum possible value. Simply, the adaptive filter is self-adjusting hence the name ’adaptive’. Below fig. 2 shows an acoustic echo cancelling setup. Let x(n) be the input signal(from the far end speaker) travelling to the near end speaker through the loud speaker and d(n) is the signal picked up by the microphone which in this case is the far end echo. h(n) represents the impulse response of the acoustic environment. w(n) represents the adaptive filter used to cancel the echo signal. The adaptive filter equate its output y(n) to the desired output d(n) ( the signal reverberated within the acoustic environment). At each iteration the error signal e(n) = d(n)-y(n) is fed back into the filter, where the filter characteristics are altered accordingly. In the case of acoustic echo cancellation, the optimal output of the adaptive filter is equal in value to the unwanted echoed signal. When the adaptive filter output is equal to desired signal the error signal goes to zero. In the situation the echoed signal is completely cancelled and the far user would hear a clear speech.
correlated output signal whereas the inverse LP predictor transforms a correlated signal back to an uncorrelated flat-spectrum signal. Inverse LP filter is an all-zero filter, with the zero situated at the same position in pole-zero plot as the poles of the all-pole filter and is also known as a spectral whitening, or decorrelation filter. Adaptive echo cancellation systems work better (i.e. converge faster) if the input and the reference signals are uncorrelated white noise processes. Speech signals are highly correlated but can be pre-whitened by first modeling the speech with a linear prediction model and then using an inverse linear predictor for whitening the signals as illustrated in fig. 3. The prewhitened input to adaptive filter, i.e. pre-whitened incoming signal, is given by e(m)= x(m)k x(m-k) A similar equation can be used to pre-whiten the adaptive echo canceller’s reference signal as shown in fig. 3. For the purpose of synthesis of the echo the input to the filter is the non-whitened speech signal, whereas for the purpose of the adaptation of the filter coefficients the whitened speech and whitened reference signals are used. The process of prewhitening the input and reference signals of the adaptive filter can substantially improve the performance of echo cancellation systems [6]. Simulink model for LMS adaptive AEC with LP and ILP Filters is designed in MATLAB. Audio files for near-end and Far-end signal are recorded as a wave files and then stored in PC to be used in this model. LP is a well known all-pole method for estimating the vocal tract signal, and there are two ways: Autocorrelation Method: The autocorrelation method produces a stable but biased solution for vocal tract for a limited size window (Hamming or Hanning). Covariance Method: This method does not guarantee the stability of the estimated filter but may produce a unbiased solution for limited size window.
Fig. 2 Block diagram of an adaptive echo cancellation system
Acoustic Echo Cancellation with Linear Prediction (LP) Filter Schematically the system can be described as in fig. 3. A linear prediction (LP) model predicts/forecasts the future values of a signal from a linear combination of its past values. A linear predictor model is an all-pole filter that models the resonance (poles) of the spectral envelope of a signal or a system. The all pole LP model shapes the spectrum of the input signal by transforming an uncorrelated excitation signal to
Fig. 3 Block diagram of the LMS adaptive AEC with LP Filter
145
International Journal of Research in Advent Technology, Vol.2, No.6, June 2014 E-ISSN: 2321-9637 3. SIMULATION SET UP AND RESULTS Acoustic Echo Cancellation using LMS adaptive algorithm The Near-end and Far-end speech are prerecorded speech. That are loaded and then linked with the model via “To wave device” blocks. Far-echo is generated using “Echo Generator” subsystem. White noise having zero mean and variance unity, Near-end and Far-echo are added and fed into the LMS filter i.e. “Desired Signal”. Following values of the parameters are used in the model simulation [7]. Audio Speech • Length of audio Speech: 10 s • Sampling rate of Audio Speech: 8000 Hz • Sample width of Audio Speech: 16 bits • Speech Data type: Double • Samples per frame of Audio Speech: 80 • Frame size: 20ms Echo Generator Subsystem • Delay of Echo Generator Subsystem: Z-400, Z-800and Z-1200 • Delay type: Samples
Fig.4Acoustic Echo Cancellation using LMS adaptive algorithm
White Noise • Mean: 0 • Variance: 1 • Source type: Gaussian • Method: Ziggurat • Inherit attributes: Output port Buffer • Convert to: Frame • Buffer size: 80
Fig.5 Far echo for LMS algorithm
LMS Filter • LMS Filter Length: 60 • Convergence step size of LMS filter: 0.002 • LMS filter Rounding mode: ceiling • LMS Over flow mode: satrrate Fig. 4 shows the simulink diagram for the Acoustic Echo Cancellation using LMS adaptive algorithm. Fig.5 shows the Far echo for LMS algorithm. Fig. 6 shows the white noise for NLMS algorithm. Fig.8 shows the error for NLMS algorithm. Fig.7 shows the desired for NLMS algorithm. Fig.9 shows the far signal for NLMS algorithm. Fig.10 shows the near signal for NLMS algorithm. Fig.11 shows the output for LMS algorithm. Fig.6 white noise for LMS algorithm
146
International Journal of Research in Advent Technology, Vol.2, No.6, June 2014 E-ISSN: 2321-9637
Fig.7 desired for LMS algorithm Fig.10 near signal for LMS algorithm
Fig.11 output for LMS algorithm Fig.8 error for LMS algorithm
4. CONCLUSION
Fig.9 far signal for LMS algorithm
In modern telecommunication systems like hands-free and teleconferencing systems, the problem arise during conversation is the creation of an acoustic echo. This problem degrades the quality of the information signal. All speech processing equipments like noise cancelling headphones and hearing aids should be able to filter different kinds of interfering signals and produce a clear sound to the listener. Currently, echo cancellation is a most interesting and challenging task in any communication system. Echo is the delayed and degraded version of original signal which travels back to its source after several reflections. It occurs when an audio source and sink operate in full duplex mode, an example of this is a hands-free loudspeaker telephone. In this situation the received signal is output through the telephone loudspeaker (audio source), this audio signal is then reverberated through the physical environment and picked up by the systems microphone (audio sink).
147
International Journal of Research in Advent Technology, Vol.2, No.6, June 2014 E-ISSN: 2321-9637 The effect is the return to the distant user of time delayed and attenuated images of their original speech signal. REFERENCES [1]. Marcel Dekker. Advances in Speech Signal Processing, 1992. [2]. Thomas Drumright; Adaptive filtering, Academic Publisher, USA, spring 1998. [3]. J. G. Proakis, "Digital Communications", 3rd edition, McGraw-Hill, 1995. [4]. S. Haykin, "Adaptive Filter Theory", PrenticeHall, 3rd Ed., 1996. [5]. ] S. Park, D. Youn and S. Park, "Acoustic interference cancellation for hands-free terminals," Digital Signal Processing, 2002. DSP 2002. 2002 14th International Conference on , vol.2, pp. 1277- 1280, 2002. [6]. S. K. M. VeeraTejaGarre, "An Acoustic Echo Cancellation System based on Adaptive Algorithms," M.S. Thesis, Dept. of Signal Processing, Blekinge Tekniska Hogskola, BTH, Sweden, October 2012. [7]. I. T. M. K. B. Homana, "Echo Cancellation using Adaptive Algorithms," Design and Technology of Electronics Packages, (SIITME) 15th International Symposium., pp. 317-321, Sept 2009.
148