DocumentCode :
2781150
Title :
Time and frequency pruning for speaker identification
Author :
Besacier, L. ; Bonastre, J.F.
Author_Institution :
LIA/CERI, Avignon, France
Volume :
2
fYear :
1998
fDate :
16-20 Aug 1998
Firstpage :
1619
Abstract :
This work is an attempt to refine decisions in speaker identification. A test utterance is divided into multiple time-frequency blocks on which a normalized likelihood score is calculated. Instead of averaging the block-likelihoods along the whole test utterance, some of them are rejected (pruning) and the final score is computed with a limited number of time-frequency blocks. The results obtained in the special case of time pruning lead the authors to experiment a joint time and frequency pruning approach. The optimal percentage of blocks pruned is learned on a tuning data set with the minimum identification error criterion. Validation of the time-frequency pruning process on 567 speakers leads to a significant error rate reduction for short training and test duration
Keywords :
maximum likelihood detection; speaker recognition; time-frequency analysis; tuning; block-likelihoods; error rate reduction; frequency pruning; speaker identification; speech recognition; time pruning; tuning data set; utterance; Covariance matrix; Loudspeakers; Read only memory; Signal processing; Speech; Testing; Time frequency analysis;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Pattern Recognition, 1998. Proceedings. Fourteenth International Conference on
Conference_Location :
Brisbane, Qld.
ISSN :
1051-4651
Print_ISBN :
0-8186-8512-3
Type :
conf
DOI :
10.1109/ICPR.1998.712026
Filename :
712026
Link To Document :
بازگشت