Continuous Speech Recognition Based on High Plausibility Regions
Identifieur interne : 000C89 ( Crin/Curation ); précédent : 000C88; suivant : 000C90Continuous Speech Recognition Based on High Plausibility Regions
Auteurs : Y. Gong ; J.-P. Haton ; F. MouriaSource :
English descriptors
Abstract
We propose a new approach to phoneme-based continuous speech recognition when a time function of plausibility of observing each phoneme (spotting result ) is given. We introduce a criterion for best sentence, based on the sum of plausibilities of individual symbols composing the sentence. Based on the idea of making use of high plausibility region to reduce computation load while keeping optimality, our method finds the most plausible sentences relating to the input speech, given the plausibility \mu_{a,n} of observing each phoneme a at each time slot n. Two optimization procedures are defined to deal with following embedded search processes : (1) find the best path connecting peaks of the plausibility functions of two successive symbols, and (2) find the best time transition slot index for two given peaks. Dynamic programming is used in these two procedures. Since the best path finding algorithm does not search slot by slot, resulting recognition system is highly efficient. Experimental results show that the method gives better recognition precision while requiring about 1/20 computing time, compared to traditional D.P. based methods. The experimental system obtained 95========percnt; sentence recognition rate on a multi-speaker test.
Links toward previous steps (curation, corpus...)
- to stream Crin, to step Corpus: Pour aller vers cette notice dans l'étape Curation :000C89
Links to Exploration step
CRIN:gong91aLe document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" wicri:score="564">Continuous Speech Recognition Based on High Plausibility Regions</title></titleStmt><publicationStmt><idno type="RBID">CRIN:gong91a</idno><date when="1991" year="1991">1991</date><idno type="wicri:Area/Crin/Corpus">000C89</idno><idno type="wicri:Area/Crin/Curation">000C89</idno><idno type="wicri:explorRef" wicri:stream="Crin" wicri:step="Curation">000C89</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en">Continuous Speech Recognition Based on High Plausibility Regions</title><author><name sortKey="Gong, Y" sort="Gong, Y" uniqKey="Gong Y" first="Y." last="Gong">Y. Gong</name></author><author><name sortKey="Haton, J P" sort="Haton, J P" uniqKey="Haton J" first="J.-P." last="Haton">J.-P. Haton</name></author><author><name sortKey="Mouria, F" sort="Mouria, F" uniqKey="Mouria F" first="F." last="Mouria">F. Mouria</name></author></analytic></biblStruct></sourceDesc></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>dynamic programming</term><term>search algorithms</term><term>speech recognition</term></keywords></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en" wicri:score="4722">We propose a new approach to phoneme-based continuous speech recognition when a time function of plausibility of observing each phoneme (spotting result ) is given. We introduce a criterion for best sentence, based on the sum of plausibilities of individual symbols composing the sentence. Based on the idea of making use of high plausibility region to reduce computation load while keeping optimality, our method finds the most plausible sentences relating to the input speech, given the plausibility \mu_{a,n} of observing each phoneme a at each time slot n. Two optimization procedures are defined to deal with following embedded search processes : (1) find the best path connecting peaks of the plausibility functions of two successive symbols, and (2) find the best time transition slot index for two given peaks. Dynamic programming is used in these two procedures. Since the best path finding algorithm does not search slot by slot, resulting recognition system is highly efficient. Experimental results show that the method gives better recognition precision while requiring about 1/20 computing time, compared to traditional D.P. based methods. The experimental system obtained 95========percnt; sentence recognition rate on a multi-speaker test.</div></front></TEI><BibTex type="inproceedings"><ref>gong91a</ref><crinnumber>91-R-241</crinnumber><category>3</category><equipe>RFIA</equipe><author><e>Gong, Y.</e><e>Haton, J.-P.</e><e>Mouria, F.</e></author><title>Continuous Speech Recognition Based on High Plausibility Regions</title><booktitle>{Proceedings IEEE International Conference on Acoustics, Speech and Signal Processing, Toronto (Canada)}</booktitle><year>1991</year><volume>1</volume><pages>725-728</pages><month>may</month><keywords><e>speech recognition</e><e>search algorithms</e><e>dynamic programming</e></keywords><abstract>We propose a new approach to phoneme-based continuous speech recognition when a time function of plausibility of observing each phoneme (spotting result ) is given. We introduce a criterion for best sentence, based on the sum of plausibilities of individual symbols composing the sentence. Based on the idea of making use of high plausibility region to reduce computation load while keeping optimality, our method finds the most plausible sentences relating to the input speech, given the plausibility \mu_{a,n} of observing each phoneme a at each time slot n. Two optimization procedures are defined to deal with following embedded search processes : (1) find the best path connecting peaks of the plausibility functions of two successive symbols, and (2) find the best time transition slot index for two given peaks. Dynamic programming is used in these two procedures. Since the best path finding algorithm does not search slot by slot, resulting recognition system is highly efficient. Experimental results show that the method gives better recognition precision while requiring about 1/20 computing time, compared to traditional D.P. based methods. The experimental system obtained 95========percnt; sentence recognition rate on a multi-speaker test.</abstract></BibTex></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Crin/Curation
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000C89 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Crin/Curation/biblio.hfd -nk 000C89 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Wicri/Lorraine
|area= InforLorV4
|flux= Crin
|étape= Curation
|type= RBID
|clé= CRIN:gong91a
|texte= Continuous Speech Recognition Based on High Plausibility Regions
}}
| This area was generated with Dilib version V0.6.33. |  |