darlington - ptive Noise sed LMS David J . Darlington,...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
ptive Noise sed LMS David J. Darlington, Douglas R. Campbell Department of Electrical and Electronic Engineering, University of Paisley, High Street, Paisley PA1 2BE, Scotland, UNITED KINGDOM. Tel: +44 141 848 3428 Fm +44 141 848 3404 E-mail: [email protected] paisley .‘ac . u’k ABSTRACT An adaptive noise cancellation scheme for speech processing is described. Adaptive filters are implemented in sub-bands, based on a model of the human cochlea. A modification to the LMS structure is introduced which guarantees stability and convergence. This .modification, a non-recursive normalisation, is used both in a wideband and in a sub-band implementation of the scheme. The effect of this normalisation is to cause the speech to be distorted, indicating that there is little benefit in using normalised LMS in a sub-band scheme, whether the applhtion uses classical or intermittent noise cancellation. - 1. SUB-BAND SPEECH ENHANCEMENT APPROACH By enhancement we mean improvement of the quality or intelligibility of the speech signal, by reduction of background noise or speech distortion, and hence improvement of the signal to noise ratio (SNR) of the contaminated speech. The enhancement of speech signals corrupted by background noise using a multi-channel, sub-band adaptive system is being investigated. The sub-band system decomposes the wideband input signals into a number of fiequency-limited signals.The use of sub-band processing for speech enhancement allows diverse processing in each sub-band depending on signal power, noise power and level of coherence between signal and noise in the two channels. Implementing a classical adaptive noise cancellation scheme in a number of fiequency-limited sub-bands also permits faster convergence of the filter coefficients due to the reduction of signal power and adaptive filter length in each sub-band. The system under development is shown in Fig.1. In this work, all sub-bands are linearly spaced in the frequency domain, although the effect of modifying the distribution of filters has also been reported by the 0-7803-3629-1/96/$5.00 @I 1996 IEEE .--e----- Fig.1: Sub-band adaptive speech enhancement system. authors[ 13. In the work reported here, the algorithm for updating the filter coefficients is modified in a fashion which has been widely reported as being useful for guaranteed convergence and stability in a non-stationary environment such as speech processing. Its performance in a sub-band environment had not been reported. 2. ADAPTIVE NOISE CANCELLATION The sub-band speech enhancement scheme described here extends that of Toner and Campbell [2] and Campbell [3]. The least mean squares (LMS) algorithm is used in an adaptive noise cancellation scheme [4] to model the differential trasfer function in a number of sub-bands, between noise signals and multiple microphones. The classical noise canceller of Fig.:! assumes that desired speech (s) is present in only
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 01/27/2010 for the course EE 4343 taught by Professor Asdfasdsas during the Spring '10 term at Aarhus Universitet.

Page1 / 4

darlington - ptive Noise sed LMS David J . Darlington,...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online