松村壮史
Non-uniform unit based HMMs for
continuous speech recognition
Abstract:A novel acoustic modeling algorithm that generates non-uniform unit HMMs to
effectively cope with spectral variations in fluent speech is proposed. The algorithm
is devised for the automatic iterative generation of long-span units for non-uniform
modeling. This generation algorithm is based on an entropy reduction criterion
using text data and a maximum likelihood criterion using speech data. The effectiveness of the non-uniform unit models is confirmed by comparing likelihood values
between long-span unit HMMs and conventional phoneme-unit HMMs. Results of
classification tests showed that the non-uniform unit HMMs provide more precise
representation than do conventional phoneme-unit HMMs, and preliminary phrase
recognition tests suggest that non-uniform unit HMMs achieve higher performance
than phoneme-unit HMMs.