Acoustic front-end optimization for large vocabulary speech recognition

In this paper we describe experiments with the acoustic front{end of our large vocabulary speech recognition system. In particular, two aspects are studied: 1) linear transforms for feature extraction and 2) the modelling of the emission probabilities. Experiments are reported on a 5000{word task of the ARPA Wall Street Journal database. For the linear transforms our main results are: Filter{bank coe cients yield a word error rate of 9.3%. A cepstral decorrelation reduces the error rate from 9.3% to 8.0%. By applying a linear discriminant analysis (LDA) a further reduction in the error rate from 8.0% to 7.1% is obtained. Recognition results are similar for a LDA applied to lter{bank outputs and to cepstral coe cients. The experiments with density modelling gave the following results: Gaussian and Laplacian densities yield similar error rates. One single vector of variances or absolute deviations outperforms density{speci c or mixture{ speci c vectors.