An on-line acoustic compensation technique for robust speech recognition

In this work we report on the use of an on-line acoustic compensation technique for robust speech recognition. With this technique acoustic mismatch between training and actual conditions is reduced through acoustic mapping. At recognition stage, observation vectors delivered by the acoustic front-end are mapped into a reference acoustic space, while input data are exploited to update the statistical parameters of the mapping. Experimental results, obtained for matched and unmatched training and testing environment conditions, show that the investigated technique tangibly improves the performance of a speaker independent speech recognizer based on hidden Markov models. Furthermore, recognition results are close to those obtained with unsupervised incremental model adaptation based on maximum likelihood linear regression.