ADAPTIVEWIENER FILTERING APPROACH FOR SPEECH ENHANCEMENT
PROBLEM DEFINITION:
Speech enhancement in the past decades has focused on the suppression of additive background noise. From a signal processing point of view additive noise is easier to deal with than convolute noise or nonlinear disturbances. Moreover, due to the bursty nature of speech, it is possible to observe the noise by itself during speech pauses, which can be of great value.
Speech enhancement is a very special case of signal estimation as speech is nonstationary, and the human ear---the final judge---does not believe in a simple mathematical error criterion. Therefore subjective measurements of intelligibility and quality are required.
METHODOLOGY:
Speech enhancement aims to improve speech quality by using various algorithms. It may sound simple, but what is ment by the word quality? It can be at least clarity and intelligibility, pleasantness, or compatibility with some other method in speech processing.
Wiener filter are rather simple andworkable, but after the estimation of the background noise, one neglects the fact that the signal isactually speech. Furthermore, the phase component of the signal is left untouched. However, this isperhaps not such a bad problem; after all, human ear is not very sensitive to phase changes. The thirdrestriction in spectral subtraction methods is the processing of the speech signal in frames, so theProceeding from one frame to another must be handled with care to avoid discontinuities.
Wiener Filtering:
Wiener filtering, spectral subtraction, subspace methods and Kalman filtering are popularly used approaches for noise reduction. In the following subsections, we discuss these methods, and study their differences and similarities
The Wiener filter obtains a least squares estimate of under stationary assumptions of speech and noise. The construction of the Wiener filter requires an estimate of the power spectrum of the clean speech and the noiseIn the Wiener filter approach, the optimal estimator is designed to minimize the mean squared error.
A Wiener filter which takes the uncertainty of signal presence into account in the noisy observation Is presented. It Is compared with the magnitude spectral subtraction, the standard Wiener filter and the estimator developed by Ephraim.
Simulations show that the level of residual noise is lower. Informal listening tests confirm this remark and reveal that the proposal approach results in significant noisereduction and provides enhanced speech with colorless residualnoise. The noise is less annoying than that of the spectralsubtraction. The complexity of the this approach iscomparable with the complexity of the spectral subtraction andlower than that of the Ephraim and Malah estimator.
.
Block diagram:
Applications:
Noise reduction in speech remains a key-point of such systems.Generally, in a moving car, speech is degraded by ambientNoises due to the engine traffic and wind and the Signal toNoise Ratio (SNR) is low. Some of them are based on the wellknownspectral subtraction approach that is suitable forenhancing speech embedded in stationary noise. In thesemethods there remains usually a level of residual, unnaturalbackground noise, called musical noise.
Conclusion:
These notes summarize results achieved in the applications of an enhancement to speech
Transmission and speech recognition.Resulting into a reliable speech recognition system.
References
[1] Y. EPHRAIM, D. MALAH, "Speech Enhancement Using a
Minimum Mean Square Error Short-Time Spectral Amplitude
Estimator", IEEE Trans. on ASSP, vol. ASSP-32, nｰ6, pp.
1109-1121, December 1984.
[2] J.S. LIM, A.V. OPPENHEIM, "Enhancement and Bandwidth
Compression of Noise Speech", Proceedings of the IEEE, vol.
67, nｰ12, pp. 1586-1604, December 1979.
[3] J.H.L. HANSEN, J.R. DELLER, "Speech Enhancement and
Quality Assessment with Applications to Robust Recognition
and Coding", Tutorial, ICASSP, May 1995.
[4] R.J. McAULAY, M.L. MALPASS, "Speech Enhancement
Using a Soft-Decision Noise Suppression Filter", IEEE Trans.
on ASSP, vol. ASSP-28, nｰ2, pp. 137-145, April 1980.
[5] O. CAPPÉ, "Elimination of the Musical Noise Phenomenon
with the Ephraim and Malah Noise Suppressor", ICASSP, pp.
345-349, 1994.