Investigations on inter-speaker variability in the feature space

Authors:
R. Haeb-Umbach
Affiliations:
Philips Res. Lab., Aachen, Germany
Venue:
ICASSP '99 Proceedings of the Acoustics, Speech, and Signal Processing, 1999. on 1999 IEEE International Conference - Volume 01
Year:
1999

Citing 0
Cited 4

Feature-dependent compensation of coders in speech recognition

Signal Processing
HMM-based techniques for speech segments extraction

Scientific Programming - Hidden Markov Models
Towards an intelligent acoustic front end for automatic speech recognition: built-in speaker normalization

EURASIP Journal on Audio, Speech, and Music Processing - Intelligent Audio, Speech, and Music Processing Applications
Signal adaptive spectral envelope estimation for robust speech recognition

Speech Communication

Quantified Score

Hi-index	0.00

Visualization

Abstract

We apply Fisher variate analysis to measure the effectiveness of speaker normalization techniques. A trace criterion, which measures the ratio of the variations due to different phonemes compared to variations due to different speakers, serves as a first assessment of a feature set without the need for recognition experiments. By using this measure and by recognition experiments we demonstrate that cepstral mean normalization also has a speaker normalization effect, in addition to the well-known channel normalization effect. Similarly vocal tract normalization (VTN) is shown to remove inter-speaker variability. For VTN we show that normalization on a per sentence basis performs better than normalization on a per speaker basis. Recognition results are given on Wall Street Journal and Hub-4 databases.