A GENERAL JOINT ADDITIVE AND CONVOLUTIVE BIAS COMPENSATION APPROACH APPLIED TO NOISY LOMBARD SPEECH RECOGNITION

In this paper, a unified approach to the acoustic mismatch problem is proposed. A maximum likelihood state-based additive bias compensation algorithm is developed for the continuous density hidden Markov model (CDHMM). Based on this technique, specific bias models in the mel ceps tral and the linear spectral domains are presented. Among these models , a new polynomial trend bias model in the mel cepstral domain is deri ved, which proved effective for Lombard speech compensation. In additi on, a joint estimation algorithm for additive and convolutive bias com pensation is proposed. This algorithm is based on applying the expecta tion maximization (EM) technique in both above-mentioned domains, in c onjunction with a parallel model combination (PMC) based transformatio n. The compensation of the dynamic (difference) coefficients in the pr oposed framework is also studied. The evaluation data base consists of a 21 confusable word vocabulary uttered by 24 speakers. Three mismatc hed versions of the data base are considered, i.e., Lombard speech, 15 dB noisy Lombard speech, and 5 dB noisy Lombard speech. The proposed techniques result in 50.9%, 74.6%, and 67.3% reduction in the performa nce difference between matched and uncompensated word error rates for the three mismatch conditions, respectively. When dynamic coefficients are considered the corresponding reductions are 46.8%, 72.4%, and 70. 9%.