8th International Conference on Spoken Language Processing

Jeju Island, Korea
October 4-8, 2004

Feature-Dependent Compensation in Speech Recognition

Ivan Brito, Nestor Becerra Yoma, Carlos Molina

University of Chile, Chile

Several mismatch conditions can be modeled as an additive bias. This bias is considered independent of the observation vectors, although this approximation is not always accurate. In this paper the dependence of the bias on the observation vectors is taken into consideration in the context of compensating the GSM coding distortion in speech recognition. However, the results presented here can easily be generalized to deal with other types of mismatch. The coding-decoding distortion is modeled here as feature-dependent. This model is employed to propose an Expectation-Maximization (EM) estimation algorithm of the coding-decoding distortion that is able to cancel the effect of GSM coder with as few as one adapting utterance. Finally, the feature-dependent adaptation can give word error rate (WER) 26% lower than the feature-independent model.

Full Paper

Bibliographic reference.  Brito, Ivan / Yoma, Nestor Becerra / Molina, Carlos (2004): "Feature-dependent compensation in speech recognition", In INTERSPEECH-2004, 2057-2060.