Environmental robustness in automatic speech recognition

Publisher Website
Google Scholar
Abstract
In this paper we report our initial efforts to make SPHINX, the CMU continuous-speech speaker-independent recognition system, robust to changes in the environment. To deal with differences in noise level and spectral tilt between closc-tcking atid desk-top microphones, we propose two novel methods based on additive corrections in the cepstral domain. In the first algorithm, the additive correction depends on the instantaneous SNR of the signal. In the second technique, EM techniques are used to bes~ match the cepstral vectors of the input utter.mces to the ensemble of codebook entries representing a standard acoustical ambience. Use of the proposed algorithms dramatically improves recognition accuracy when the system is tested on a microphone other than the one on which it was trained. plicitly. In this paper we present two algorithms for speech normalization based on additive corrections in the cepstral domain. We have chosen the cepstral domain rather than the frequency domain so that we work directly with the parameters that SPHINX uses, and because speech can be characterized with a smaller number of parameters in the cepstral domain than in the frequency domain. The first algorithm, SNR-deperidenf cepstral normalization (SDCN) is simple and effective, but it cannot be applied to new microphones without microphone-specific training. The second algorithm, codeword-deperident cepstral norntalizution (CDCN) computes an ML estimate for the noise and spectral tilt, and then an MMSE estimate for the speech cepstrum. These algorithms are evaluated using an alphanumeric database in which utterances were recorded simultaneously with two different microphones.

This publication has 7 references indexed in Scilit: