http://cslt.org/mediawiki/index.php?action=history&feed=atom&title=Hulan-2013-09-27 2026-04-15T20:06:58Z 本wiki的该页面的版本历史 MediaWiki 1.23.3 http://cslt.org/mediawiki/index.php?title=Hulan-2013-09-27&diff=8231&oldid=prev 2013-09-27T01:36:33Z

<p></p> <table class='diff diff-contentalign-left'> <col class='diff-marker' /> <col class='diff-content' /> <col class='diff-marker' /> <col class='diff-content' /> <tr style='vertical-align: top;'> <td colspan='2' style="background-color: white; color:black; text-align: center;">←上一版本</td> <td colspan='2' style="background-color: white; color:black; text-align: center;">2013年9月27日 (五) 01:36的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第7行：</td> <td colspan="2" class="diff-lineno">第7行：</td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"><div>==TTS==</div></td><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"><div>==TTS==</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"></td><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"></td></tr> <tr><td class='diff-marker'>−</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #ffe49c; vertical-align: top; white-space: pre-wrap;"><div>* <del class="diffchange diffchange-inline">EST </del>format learned<del class="diffchange diffchange-inline">.</del></div></td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div>* <ins class="diffchange diffchange-inline">Lab </ins>format learned</div></td></tr> <tr><td class='diff-marker'>−</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #ffe49c; vertical-align: top; white-space: pre-wrap;"><div>* <del class="diffchange diffchange-inline">Check </del>details of <del class="diffchange diffchange-inline">each option</del>.</div></td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div>* <ins class="diffchange diffchange-inline">All the </ins>details of <ins class="diffchange diffchange-inline">the label format are clear</ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div><ins class="diffchange diffchange-inline">* Construct label files from word/pingyin/phone transcription</ins>. <ins class="diffchange diffchange-inline">Use csep word-segmentation tool to obtain these transcriptions from the original text.</ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div><ins class="diffchange diffchange-inline">* Monophone, Triphone Chinese prototype system is ready. 500 sentences from 863 data are used for training. The trivial questions were used for clustering. 16k signals with 256 FFT transform. GV model used.</ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div><ins class="diffchange diffchange-inline">* The voice is funny. </ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div>&#160;</div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div><ins class="diffchange diffchange-inline">Next week:</ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div><ins class="diffchange diffchange-inline">* Keep on collecting context-dependent labels</ins></div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div>&#160;</div></td></tr> <tr><td colspan="2">&#160;</td><td class='diff-marker'>+</td><td style="color:black; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #a3d3ff; vertical-align: top; white-space: pre-wrap;"><div>&#160;</div></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"></td><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"></td></tr> <tr><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"><div>=Dialog system=</div></td><td class='diff-marker'>&#160;</td><td style="background-color: #f9f9f9; color: #333333; font-size: 88%; border-style: solid; border-width: 1px 1px 1px 4px; border-radius: 0.33em; border-color: #e6e6e6; vertical-align: top; white-space: pre-wrap;"><div>=Dialog system=</div></td></tr> </table>

Cslt http://cslt.org/mediawiki/index.php?title=Hulan-2013-09-27&diff=8230&oldid=prev 2013-09-27T01:25:18Z

<p>以内容“=ASR= ==ASR Kernel development== <a href="/mediawiki/index.php?title=Http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/2013-09-27_ASR_group_weekly_report&amp;action=edit&amp;redlink=1" class="new" title="Http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/2013-09-27 ASR group weekly report（页面不存在）">http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/2013-09-27 ASR group weekly report</a> ==TTS== * EST format learned. * Check ...”创建新页面</p> <p><b>新页面</b></p><div>=ASR=<br /> <br /> ==ASR Kernel development==<br /> <br /> [[http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/2013-09-27 ASR group weekly report]]<br /> <br /> ==TTS==<br /> <br /> * EST format learned.<br /> * Check details of each option.<br /> <br /> =Dialog system=<br /> <br /> * MH, RY help compose 60 questions, which are being used for testing.<br /> * Conducting the initial experiment:<br /> <br /> #Using 9k dim TF/IDF, compose feature vectors for each query, each answer. Mach the TF/IDF of query+answer to match the TF/IDF of new queries. Add the scores of the Cosine score of the match with queries and answers directly.<br /> <br /> * CER: 8.3%<br /> * Query time: very slow<br /> <br /> # Remove 506 stop words. No significant change on CER &amp; Query time.<br /> <br /> # Fast match by listing the words only in the query &amp; answer as the feature, and matching by order to speed up the score calculation. <br /> <br /> # Hierarchical matching. First split all the answers to 11 top-level categories, and then split them into 1030 second-level categories. query+answer TF/IDF score.<br /> <br /> * CER: top-category: 6.7%, second-level category: 18.3%<br /> * Query time: 6/sec<br /> <br /> # Keep two top-level categories, try to reduce top-level errors:<br /> <br /> * CER: two top-category, no errors. second-level category: still on going<br /> *<br /> <br /> Next week: <br /> <br /> * Reverse index-based fast match, on going.<br /> *</div>

Cslt