Efficient search with posterior probability estimates in HMM-based speech recognition
In this paper we present the methods we developed to estimate posterior probabilities for HMM states in continuous and discrete HMM-based speech recognition systems and several ways to speed up decoding by using these posterior probability estimates. The proposed pruning techniques are state deactiv...
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Tagungsbericht |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 824 vol.2 |
|---|---|
| container_issue | |
| container_start_page | 821 |
| container_title | |
| container_volume | 2 |
| creator | Willett, D. Neukirchen, C. Rigoll, G. |
| description | In this paper we present the methods we developed to estimate posterior probabilities for HMM states in continuous and discrete HMM-based speech recognition systems and several ways to speed up decoding by using these posterior probability estimates. The proposed pruning techniques are state deactivation pruning (SDP), similar to an approach proposed for hybrid recognition systems, and a novel posteriori-based lookahead technique, posteriori lookahead pruning (PLP), that evaluates future posteriors in order to exclude unlikely HMM states as early as possible during search. By applying the proposed methods we managed to vastly reduce the decoding time consumed by our time-synchronous Viterbi-decoder for recognition systems based on the Verbmobil and the Wall Street Journal database with hardly any additional search error. |
| doi_str_mv | 10.1109/ICASSP.1998.675391 |
| format | Conference Proceeding |
| fullrecord | <record><control><sourceid>ieee_6IE</sourceid><recordid>TN_cdi_ieee_primary_675391</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>675391</ieee_id><sourcerecordid>675391</sourcerecordid><originalsourceid>FETCH-LOGICAL-i174t-575d6b63fbb9247b18c4f9310d4ee4390b2be967efc1b6b0d82e5ff9a943d74c3</originalsourceid><addsrcrecordid>eNotkE1qwzAUhEV_oCb1BbLSBZxKlmT5LUtIm0JCC2mguyDJT41KahtJUHL7GpJhYFYzfAwhc84WnDN4els-73YfCw7QLhqtBPAbUtRCQ8WBfd2SEnTLJgsp65bdkYKrmlUNl_BAypR-2CSpFNOqIPuV98EF7DNNaKI70r-Qj3QcUsYYhkjHOFhjwynkM8WUw6_JmGjo6Xq7raxJ2NE0Ik7FiG747kMOQ_9I7r05JSyvOSP7l9Xncl1t3l8n-k0VuJa5Ulp1jW2EtxZqqS1vnfQgOOskohTAbG0RGo3ecdtY1rU1Ku_BgBSdlk7MyPyyGxDxMMaJLp4Pl0_EP_7mVQ8</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Efficient search with posterior probability estimates in HMM-based speech recognition</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Willett, D. ; Neukirchen, C. ; Rigoll, G.</creator><creatorcontrib>Willett, D. ; Neukirchen, C. ; Rigoll, G.</creatorcontrib><description>In this paper we present the methods we developed to estimate posterior probabilities for HMM states in continuous and discrete HMM-based speech recognition systems and several ways to speed up decoding by using these posterior probability estimates. The proposed pruning techniques are state deactivation pruning (SDP), similar to an approach proposed for hybrid recognition systems, and a novel posteriori-based lookahead technique, posteriori lookahead pruning (PLP), that evaluates future posteriors in order to exclude unlikely HMM states as early as possible during search. By applying the proposed methods we managed to vastly reduce the decoding time consumed by our time-synchronous Viterbi-decoder for recognition systems based on the Verbmobil and the Wall Street Journal database with hardly any additional search error.</description><identifier>ISSN: 1520-6149</identifier><identifier>ISBN: 9780780344280</identifier><identifier>ISBN: 0780344286</identifier><identifier>EISSN: 2379-190X</identifier><identifier>DOI: 10.1109/ICASSP.1998.675391</identifier><language>eng</language><publisher>IEEE</publisher><subject>Computer science ; Context modeling ; Decoding ; Distribution functions ; Hidden Markov models ; Laplace equations ; Neural networks ; Speech recognition ; State estimation</subject><ispartof>Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181), 1998, Vol.2, p.821-824 vol.2</ispartof><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/675391$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,778,782,787,788,2054,4038,4039,27908,54903</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/675391$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Willett, D.</creatorcontrib><creatorcontrib>Neukirchen, C.</creatorcontrib><creatorcontrib>Rigoll, G.</creatorcontrib><title>Efficient search with posterior probability estimates in HMM-based speech recognition</title><title>Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181)</title><addtitle>ICASSP</addtitle><description>In this paper we present the methods we developed to estimate posterior probabilities for HMM states in continuous and discrete HMM-based speech recognition systems and several ways to speed up decoding by using these posterior probability estimates. The proposed pruning techniques are state deactivation pruning (SDP), similar to an approach proposed for hybrid recognition systems, and a novel posteriori-based lookahead technique, posteriori lookahead pruning (PLP), that evaluates future posteriors in order to exclude unlikely HMM states as early as possible during search. By applying the proposed methods we managed to vastly reduce the decoding time consumed by our time-synchronous Viterbi-decoder for recognition systems based on the Verbmobil and the Wall Street Journal database with hardly any additional search error.</description><subject>Computer science</subject><subject>Context modeling</subject><subject>Decoding</subject><subject>Distribution functions</subject><subject>Hidden Markov models</subject><subject>Laplace equations</subject><subject>Neural networks</subject><subject>Speech recognition</subject><subject>State estimation</subject><issn>1520-6149</issn><issn>2379-190X</issn><isbn>9780780344280</isbn><isbn>0780344286</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>1998</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNotkE1qwzAUhEV_oCb1BbLSBZxKlmT5LUtIm0JCC2mguyDJT41KahtJUHL7GpJhYFYzfAwhc84WnDN4els-73YfCw7QLhqtBPAbUtRCQ8WBfd2SEnTLJgsp65bdkYKrmlUNl_BAypR-2CSpFNOqIPuV98EF7DNNaKI70r-Qj3QcUsYYhkjHOFhjwynkM8WUw6_JmGjo6Xq7raxJ2NE0Ik7FiG747kMOQ_9I7r05JSyvOSP7l9Xncl1t3l8n-k0VuJa5Ulp1jW2EtxZqqS1vnfQgOOskohTAbG0RGo3ecdtY1rU1Ku_BgBSdlk7MyPyyGxDxMMaJLp4Pl0_EP_7mVQ8</recordid><startdate>1998</startdate><enddate>1998</enddate><creator>Willett, D.</creator><creator>Neukirchen, C.</creator><creator>Rigoll, G.</creator><general>IEEE</general><scope>6IE</scope><scope>6IH</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIO</scope></search><sort><creationdate>1998</creationdate><title>Efficient search with posterior probability estimates in HMM-based speech recognition</title><author>Willett, D. ; Neukirchen, C. ; Rigoll, G.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i174t-575d6b63fbb9247b18c4f9310d4ee4390b2be967efc1b6b0d82e5ff9a943d74c3</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>1998</creationdate><topic>Computer science</topic><topic>Context modeling</topic><topic>Decoding</topic><topic>Distribution functions</topic><topic>Hidden Markov models</topic><topic>Laplace equations</topic><topic>Neural networks</topic><topic>Speech recognition</topic><topic>State estimation</topic><toplevel>online_resources</toplevel><creatorcontrib>Willett, D.</creatorcontrib><creatorcontrib>Neukirchen, C.</creatorcontrib><creatorcontrib>Rigoll, G.</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan (POP) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Electronic Library (IEL)</collection><collection>IEEE Proceedings Order Plans (POP) 1998-present</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Willett, D.</au><au>Neukirchen, C.</au><au>Rigoll, G.</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Efficient search with posterior probability estimates in HMM-based speech recognition</atitle><btitle>Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181)</btitle><stitle>ICASSP</stitle><date>1998</date><risdate>1998</risdate><volume>2</volume><spage>821</spage><epage>824 vol.2</epage><pages>821-824 vol.2</pages><issn>1520-6149</issn><eissn>2379-190X</eissn><isbn>9780780344280</isbn><isbn>0780344286</isbn><abstract>In this paper we present the methods we developed to estimate posterior probabilities for HMM states in continuous and discrete HMM-based speech recognition systems and several ways to speed up decoding by using these posterior probability estimates. The proposed pruning techniques are state deactivation pruning (SDP), similar to an approach proposed for hybrid recognition systems, and a novel posteriori-based lookahead technique, posteriori lookahead pruning (PLP), that evaluates future posteriors in order to exclude unlikely HMM states as early as possible during search. By applying the proposed methods we managed to vastly reduce the decoding time consumed by our time-synchronous Viterbi-decoder for recognition systems based on the Verbmobil and the Wall Street Journal database with hardly any additional search error.</abstract><pub>IEEE</pub><doi>10.1109/ICASSP.1998.675391</doi></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISSN: 1520-6149 |
| ispartof | Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181), 1998, Vol.2, p.821-824 vol.2 |
| issn | 1520-6149 2379-190X |
| language | eng |
| recordid | cdi_ieee_primary_675391 |
| source | IEEE Electronic Library (IEL) Conference Proceedings |
| subjects | Computer science Context modeling Decoding Distribution functions Hidden Markov models Laplace equations Neural networks Speech recognition State estimation |
| title | Efficient search with posterior probability estimates in HMM-based speech recognition |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-16T11%3A20%3A59IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-ieee_6IE&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Efficient%20search%20with%20posterior%20probability%20estimates%20in%20HMM-based%20speech%20recognition&rft.btitle=Proceedings%20of%20the%201998%20IEEE%20International%20Conference%20on%20Acoustics,%20Speech%20and%20Signal%20Processing,%20ICASSP%20'98%20(Cat.%20No.98CH36181)&rft.au=Willett,%20D.&rft.date=1998&rft.volume=2&rft.spage=821&rft.epage=824%20vol.2&rft.pages=821-824%20vol.2&rft.issn=1520-6149&rft.eissn=2379-190X&rft.isbn=9780780344280&rft.isbn_list=0780344286&rft_id=info:doi/10.1109/ICASSP.1998.675391&rft_dat=%3Cieee_6IE%3E675391%3C/ieee_6IE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=675391&rfr_iscdi=true |