Acoustic front-end optimization for large vocabulary speech recognition
Lutz Welling, Nils Haberland, Hermann Ney · 1997
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.