Generation and expansion of word graphs using long span context information
An algorithm for the generation of word graphs in a cross-word decoder that uses long span m-gram language models (LMs) is presented. The generation of word hypotheses within the graph relies on the word m-tuple-based boundary optimization. The graphs contain the full word history knowledge informat...
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 | 44 vol.1 |
|---|---|
| container_issue | |
| container_start_page | 41 |
| container_title | |
| container_volume | 1 |
| creator | Neukirchen, C. Klakow, D. Aubert, X. |
| description | An algorithm for the generation of word graphs in a cross-word decoder that uses long span m-gram language models (LMs) is presented. The generation of word hypotheses within the graph relies on the word m-tuple-based boundary optimization. The graphs contain the full word history knowledge information since the graph structure reflects all LM constraints used during the search. This results in better word boundaries and in enhanced capabilities to prune the graphs. Furthermore, the memory costs for expanding the m-gram constrained word graphs to apply very long span LMs (e.g. ten-grams that are constructed by log linear LM combination) are considerably reduced. Experiments for lattice generation and rescoring have been carried out on the 5K-word WSJ task and the 64K-word NAB task. |
| doi_str_mv | 10.1109/ICASSP.2001.940762 |
| format | Conference Proceeding |
| fullrecord | <record><control><sourceid>ieee_6IE</sourceid><recordid>TN_cdi_ieee_primary_940762</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>940762</ieee_id><sourcerecordid>940762</sourcerecordid><originalsourceid>FETCH-LOGICAL-i87t-1438b53c005a27fc87caa886126d956321a7e0daa947facc72ba34014e3932e73</originalsourceid><addsrcrecordid>eNotkNtKAzEYhIMHcFt9gV7lBXb9c9hNcilFq1hQaC-8K3-z2RppkyVZsb69q_VmhoHhYxhCZgwqxsDcPs3vVqvXigOwykhQDT8jBRfKlMzA2zmZgNIgFEgmL0jBag5lw6S5IpOcPwBAK6kL8rxwwSUcfAwUQ0vdsceQf1Ps6FdMLd0l7N8z_cw-7Og-jpLHCrUxDO44UB-6mA5_gGty2eE-u5t_n5L1w_16_lguXxbj2mXptRpKJoXe1sIC1MhVZ7WyiFo3jDetqRvBGSoHLaKRqkNrFd-ikMCkE0Zwp8SUzE5Y75zb9MkfMH1vTheIH0JTTwA</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Generation and expansion of word graphs using long span context information</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Neukirchen, C. ; Klakow, D. ; Aubert, X.</creator><creatorcontrib>Neukirchen, C. ; Klakow, D. ; Aubert, X.</creatorcontrib><description>An algorithm for the generation of word graphs in a cross-word decoder that uses long span m-gram language models (LMs) is presented. The generation of word hypotheses within the graph relies on the word m-tuple-based boundary optimization. The graphs contain the full word history knowledge information since the graph structure reflects all LM constraints used during the search. This results in better word boundaries and in enhanced capabilities to prune the graphs. Furthermore, the memory costs for expanding the m-gram constrained word graphs to apply very long span LMs (e.g. ten-grams that are constructed by log linear LM combination) are considerably reduced. Experiments for lattice generation and rescoring have been carried out on the 5K-word WSJ task and the 64K-word NAB task.</description><identifier>ISSN: 1520-6149</identifier><identifier>ISBN: 0780370414</identifier><identifier>ISBN: 9780780370418</identifier><identifier>EISSN: 2379-190X</identifier><identifier>DOI: 10.1109/ICASSP.2001.940762</identifier><language>eng</language><publisher>IEEE</publisher><subject>Acoustics ; Context modeling ; Costs ; Decoding ; Gold ; Hidden Markov models ; History ; Laboratories ; Speech recognition ; Tree graphs</subject><ispartof>2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221), 2001, Vol.1, p.41-44 vol.1</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/940762$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,2058,4050,4051,27925,54920</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/940762$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Neukirchen, C.</creatorcontrib><creatorcontrib>Klakow, D.</creatorcontrib><creatorcontrib>Aubert, X.</creatorcontrib><title>Generation and expansion of word graphs using long span context information</title><title>2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221)</title><addtitle>ICASSP</addtitle><description>An algorithm for the generation of word graphs in a cross-word decoder that uses long span m-gram language models (LMs) is presented. The generation of word hypotheses within the graph relies on the word m-tuple-based boundary optimization. The graphs contain the full word history knowledge information since the graph structure reflects all LM constraints used during the search. This results in better word boundaries and in enhanced capabilities to prune the graphs. Furthermore, the memory costs for expanding the m-gram constrained word graphs to apply very long span LMs (e.g. ten-grams that are constructed by log linear LM combination) are considerably reduced. Experiments for lattice generation and rescoring have been carried out on the 5K-word WSJ task and the 64K-word NAB task.</description><subject>Acoustics</subject><subject>Context modeling</subject><subject>Costs</subject><subject>Decoding</subject><subject>Gold</subject><subject>Hidden Markov models</subject><subject>History</subject><subject>Laboratories</subject><subject>Speech recognition</subject><subject>Tree graphs</subject><issn>1520-6149</issn><issn>2379-190X</issn><isbn>0780370414</isbn><isbn>9780780370418</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2001</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><sourceid>RIE</sourceid><recordid>eNotkNtKAzEYhIMHcFt9gV7lBXb9c9hNcilFq1hQaC-8K3-z2RppkyVZsb69q_VmhoHhYxhCZgwqxsDcPs3vVqvXigOwykhQDT8jBRfKlMzA2zmZgNIgFEgmL0jBag5lw6S5IpOcPwBAK6kL8rxwwSUcfAwUQ0vdsceQf1Ps6FdMLd0l7N8z_cw-7Og-jpLHCrUxDO44UB-6mA5_gGty2eE-u5t_n5L1w_16_lguXxbj2mXptRpKJoXe1sIC1MhVZ7WyiFo3jDetqRvBGSoHLaKRqkNrFd-ikMCkE0Zwp8SUzE5Y75zb9MkfMH1vTheIH0JTTwA</recordid><startdate>2001</startdate><enddate>2001</enddate><creator>Neukirchen, C.</creator><creator>Klakow, D.</creator><creator>Aubert, X.</creator><general>IEEE</general><scope>6IE</scope><scope>6IH</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIO</scope></search><sort><creationdate>2001</creationdate><title>Generation and expansion of word graphs using long span context information</title><author>Neukirchen, C. ; Klakow, D. ; Aubert, X.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i87t-1438b53c005a27fc87caa886126d956321a7e0daa947facc72ba34014e3932e73</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2001</creationdate><topic>Acoustics</topic><topic>Context modeling</topic><topic>Costs</topic><topic>Decoding</topic><topic>Gold</topic><topic>Hidden Markov models</topic><topic>History</topic><topic>Laboratories</topic><topic>Speech recognition</topic><topic>Tree graphs</topic><toplevel>online_resources</toplevel><creatorcontrib>Neukirchen, C.</creatorcontrib><creatorcontrib>Klakow, D.</creatorcontrib><creatorcontrib>Aubert, X.</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>Neukirchen, C.</au><au>Klakow, D.</au><au>Aubert, X.</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Generation and expansion of word graphs using long span context information</atitle><btitle>2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221)</btitle><stitle>ICASSP</stitle><date>2001</date><risdate>2001</risdate><volume>1</volume><spage>41</spage><epage>44 vol.1</epage><pages>41-44 vol.1</pages><issn>1520-6149</issn><eissn>2379-190X</eissn><isbn>0780370414</isbn><isbn>9780780370418</isbn><abstract>An algorithm for the generation of word graphs in a cross-word decoder that uses long span m-gram language models (LMs) is presented. The generation of word hypotheses within the graph relies on the word m-tuple-based boundary optimization. The graphs contain the full word history knowledge information since the graph structure reflects all LM constraints used during the search. This results in better word boundaries and in enhanced capabilities to prune the graphs. Furthermore, the memory costs for expanding the m-gram constrained word graphs to apply very long span LMs (e.g. ten-grams that are constructed by log linear LM combination) are considerably reduced. Experiments for lattice generation and rescoring have been carried out on the 5K-word WSJ task and the 64K-word NAB task.</abstract><pub>IEEE</pub><doi>10.1109/ICASSP.2001.940762</doi></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISSN: 1520-6149 |
| ispartof | 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221), 2001, Vol.1, p.41-44 vol.1 |
| issn | 1520-6149 2379-190X |
| language | eng |
| recordid | cdi_ieee_primary_940762 |
| source | IEEE Electronic Library (IEL) Conference Proceedings |
| subjects | Acoustics Context modeling Costs Decoding Gold Hidden Markov models History Laboratories Speech recognition Tree graphs |
| title | Generation and expansion of word graphs using long span context information |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-29T05%3A29%3A00IST&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=Generation%20and%20expansion%20of%20word%20graphs%20using%20long%20span%20context%20information&rft.btitle=2001%20IEEE%20International%20Conference%20on%20Acoustics,%20Speech,%20and%20Signal%20Processing.%20Proceedings%20(Cat.%20No.01CH37221)&rft.au=Neukirchen,%20C.&rft.date=2001&rft.volume=1&rft.spage=41&rft.epage=44%20vol.1&rft.pages=41-44%20vol.1&rft.issn=1520-6149&rft.eissn=2379-190X&rft.isbn=0780370414&rft.isbn_list=9780780370418&rft_id=info:doi/10.1109/ICASSP.2001.940762&rft_dat=%3Cieee_6IE%3E940762%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=940762&rfr_iscdi=true |