NOISE COMPENSATION METHODS FOR HIDDEN MARKOV MODEL SPEECH RECOGNITIONIN ADVERSE ENVIRONMENTS

Several noise compensation schemes for speech recognition in impulsive and nonimpulsive noise are considered. The noise compensation schemes are spectral subtraction, HMM-based Wiener filters, noise-adaptive HM M's, and a front-end impulsive noise removal. The use of the cepstral- time matrix as an improved speech feature set is explored, and the noi se compensation methods are extended for use with cepstral-time featur es. Experimental evaluations, on a spoken digit database, in the prese nce of car noise, helicopter noise, and impulsive noise, demonstrate t hat the noise compensation methods achieve substantial improvement in recognition across a wide range of signal-to-noise ratios. The results also show that the Cepstral-time matrix is more robust than a vector of identical size, which is composed of a combination of cepstral and differential cepstral features.