BAYESIAN ENHANCEMENT OF SPEECH AND AUDIO SIGNALS WHICH CAN BE MODELEDAS ARMA PROCESSES

In application areas which involve digitised speech and audio signals, such as coding, digital remastering of old recordings and recognition of speech, it is often desirable to reduce the effects of noise with the aim of enhancing intelligibility and perceived sound quality, We c onsider the case where noise sources contain non-Gaussian, impulsive e lements superimposed upon a continuous Gaussian background. Such a sit uation arises in areas such as communications channels, telephony and gramophone recordings where impulsive effects might be caused by elect romagnetic interference (lightning strikes), electrical switching nois e or defects in recording media, while electrical circuit noise or the combined effect of many distant atmospheric events lead to a continuo us Gaussian component. In this paper we discuss the background to this type of noise degradation and describe briefly some existing statisti cal techniques for noise reduction, We propose new methods for enhance ment based upon Markov chain Monte Carlo (MCMC) simulation. Signals ar e modelled as autoregressive moving-average (ARMA); while noise source s are treated as discrete and continuous mixtures of Gaussian distribu tions. Results are presented for both real and artificially corrupted data sequences, illustrating the potential of the new methods.