http://index.cslt.org/mediawiki/index.php?action=history&feed=atom&title=Xinsong-beamforming-result 2026-04-07T12:12:32Z 本wiki的该页面的版本历史 MediaWiki 1.23.3 http://index.cslt.org/mediawiki/index.php?title=Xinsong-beamforming-result&diff=20222&oldid=prev 2016-05-12T03:06:17Z

<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;">2016年5月12日 (四) 03:06的版本</td> </tr><tr><td colspan="2" class="diff-lineno">第8行：</td> <td colspan="2" class="diff-lineno">第8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>ID3: test data which are involved in the DAE training &#160;</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>ID3: test data which are involved in the DAE training &#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 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 style="font-weight: bold; text-decoration: none;">&lt;pre&gt;</ins></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;"><div>baseline:</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>baseline:</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;"><div>---------------------------------------------------</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>---------------------------------------------------</div></td></tr> <tr><td colspan="2" class="diff-lineno">第117行：</td> <td colspan="2" class="diff-lineno">第118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>_tdnn_lr_0.008_c4 |</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>_tdnn_lr_0.008_c4 |</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;"><div>-------------------------------------------------------</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>-------------------------------------------------------</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 style="font-weight: bold; text-decoration: none;"></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 style="font-weight: bold; text-decoration: none;">&lt;/pre&gt;</ins></div></td></tr> </table>

Cslt http://index.cslt.org/mediawiki/index.php?title=Xinsong-beamforming-result&diff=20221&oldid=prev 2016-05-12T03:05:53Z

<p>以“far field SNR : 14dB restaurant training data reorded in 2016.4.22 has 250 sentences and test data has 11 sentences. 16k and 16bit data AM: 10000h 7*2048 MPE: LM:...”为内容创建页面</p> <p><b>新页面</b></p><div>far field SNR : 14dB<br /> restaurant training data reorded in 2016.4.22 has 250 sentences and test data has 11 sentences.<br /> 16k and 16bit data<br /> <br /> AM: 10000h 7*2048 MPE:<br /> LM: 1e-7 5-gram<br /> <br /> ID3: test data which are involved in the DAE training <br /> <br /> baseline:<br /> ---------------------------------------------------<br /> | test random wer | test ID3 wer(within trainingset)<br /> ---------------------------------------------------<br /> c1 | 39.74 | 36.89<br /> ---------------------------------------------------<br /> c2 | 30.77 | 34.95<br /> ---------------------------------------------------<br /> c3 | 34.62 | 37.38<br /> ---------------------------------------------------<br /> c4 | 42.31 | 36.89<br /> ---------------------------------------------------<br /> near | 21.79 | 4.37<br /> ---------------------------------------------------<br /> <br /> beamforming:<br /> ---------------------------------------------------<br /> DS_post | 26.92 | 29.13<br /> ----------------------------------------------------<br /> SD_post | 23.08 | 26.70 (second)<br /> ----------------------------------------------------<br /> MVDR_post | 26.92 | 28.64<br /> ----------------------------------------------------<br /> sino beamforming|<br /> _from_xiaoming | 26.92 | 31.07 <br /> ----------------------------------------------------<br /> <br /> Four-channel cnn_tdnn model: 160 fbank <br /> <br /> ----------------------------------------------------<br /> tdnn_dae_1*1024| 38.46 | 10.19<br /> _tdnn_lr_0.008 |<br /> -----------------------------------------------------<br /> cnn_tdnn_daa_ |<br /> 1*128_cnn_1*1024| 39.74 | 5.34<br /> _tdnn_lr_0.008 |<br /> -----------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 1*128_cnn_1*512 | 48.72 | 7.28<br /> _tdnn_lr_0.008 |<br /> -----------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 1*64_cnn_1*1024 | 39.74 | 7.28<br /> _tdnn_lr_0.008 |<br /> -----------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 2*64_cnn_1*1024 | 50.00 | 25.24 <br /> _tdnn_lr_0.008 |<br /> -----------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 2*64_cnn_1*1024 | 44.87 | 23.30 <br /> _tdnn_lr_0.008_ |<br /> nopooling |<br /> -----------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 1*32_cnn_1*1024 | 50.00 | 8.25<br /> _tdnn_lr_0.008 |<br /> ------------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 2*32_cnn_1*1024 | 46.15 | 33.98 <br /> _tdnn_lr_0.008 |<br /> ------------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 2*32_cnn_1*1024 | 47.44 | 28.16 <br /> _tdnn_lr_0.008_ |<br /> nopooling |<br /> ------------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 1*16_cnn_1*1024 | 46.15 | 8.25 <br /> _tdnn_lr_0.008 |<br /> -------------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 2*16_cnn_1*1024 | 62.82 | 17.96<br /> _tdnn_lr_0.008 |<br /> -------------------------------------------------<br /> cnn_tdnn_dae_ | <br /> 2*16_cnn_1*1024 | 50.00 | 22.82 <br /> _tdnn_lr_0.008_ |<br /> nopooling |<br /> -------------------------------------------------<br /> cnn_tdnn_dae_ |<br /> 2*16_cnn_1*256 | 48.72 | 16.99 <br /> _tdnn_lr_0.008 |<br /> -------------------------------------------------<br /> cnn_tdnn_dae_ | <br /> 2*16_cnn_1*256 | 48.72 | 19.90<br /> _tdnn_lr_0.008_ |<br /> nopooling |<br /> -------------------------------------------------<br /> <br /> <br /> Single channel cnn_tdnn: 40 Fbanks<br /> ------------------------------------------------------<br /> cnn_tdnn_mix_dae_| <br /> 1*128_cnn_1*1024 | 25.64 | 9.22 <br /> _tdnn_lr_0.008_c1 |<br /> ------------------------------------------------------<br /> cnn_tdnn_mix_dae_| <br /> 1*128_cnn_1*1024 | 28.21 | 7.28<br /> _tdnn_lr_0.008_c2 |<br /> ------------------------------------------------------<br /> cnn_tdnn_mix_dae_ | <br /> 1*128_cnn_1*1024 | 20.51 | 4.37 (first)<br /> _tdnn_lr_0.008_c3 |<br /> ------------------------------------------------------<br /> cnn_tdnn_mix_dae_ | <br /> 1*128_cnn_1*1024 | 21.79 | 5.34<br /> _tdnn_lr_0.008_c4 |<br /> -------------------------------------------------------</div>

Cslt