Reducing the Effects of Linear Channel Distortion on Continuous Speech Recognition
Linear channel compensation in speech recognition typically involves estimating an additive shift in the cepstral domain. This paper explores both Bayesian and maximum likelihood techniques to transform either the features or the model parameters. Experiments on the Macrophone corpus show error rate reductions of up to 16% over cepstral mean subtraction for short utterances.
IEEE Transactions on Speech and Audio Processing
Bates, R., & Ostendorf, M. (1999). Reducing the Effects of Linear Channel distortion on Continuous Speech Recognition. IEEE Transactions on Speech and Audio Processing, 7(5), 594-597. doi:10.1109/89.784112
Publisher's Copyright and Source
Copyright © 1999 IEEE. Article published by IEEE in IEEE Transactions on Speech and Audio Processing, volume 7, issue number 5, September 1999, pages 594-597. Available online at https://doi.org/10.1109/89.784112.