http://cslt.org/mediawiki/index.php?action=history&feed=atom&title=2014-04-18 2026-04-14T11:24:10Z 本wiki的该页面的版本历史 MediaWiki 1.23.3 http://cslt.org/mediawiki/index.php?title=2014-04-18&diff=9727&oldid=prev 2014-04-18T02:35:00Z

<p>以内容“==Resoruce Building== * quota on /nfs/disk this Saturday * release management should be started: Zhiyong * Blaster 0.1 &amp; vivian 0.0 system release == Leftover question...”创建新页面</p> <p><b>新页面</b></p><div>==Resoruce Building==<br /> * quota on /nfs/disk this Saturday<br /> * release management should be started: Zhiyong<br /> * Blaster 0.1 &amp; vivian 0.0 system release<br /> <br /> == Leftover questions==<br /> * Asymmetric window: Great improvement on training set(WER 34% to 24%), however the improvement is lost on test. Overfitting? <br /> * Multi GPU training: Error encountered<br /> * Multilanguage training<br /> * Investigating LOUDS FST. <br /> * CLG embedded decoder plus online compiler.<br /> * DNN-GMM co-training<br /> <br /> == AM development ==<br /> <br /> === Sparse DNN ===<br /> * GA-based block sparsity<br /> :* Found a paper in 2000 with similar ideas. <br /> :* Try to get a student working on high performance computing to do the optimization<br /> <br /> ===Noise training===<br /> :* More experiments with no-noise<br /> :* More experiments with additional noise types<br /> <br /> ===AMR compression re-training===<br /> <br /> * 1700h MPE adaptation done<br /> * 1700h stream mode adaptation runs into MPE4 done<br /> * Stream model is better than non-stream wave<br /> <br /> ===GFbank===<br /> * GFBank Sinovoice test on 100h MPE<br /> * Tencent 100h MPE training done<br /> <br /> ===Multilingual ASR===<br /> * all phone strategy baseline done<br /> * Testing on Mandarin &amp; English individually<br /> <br /> ===Denoising &amp; Farfield ASR===<br /> * re-Recording done<br /> * Prepare to construct the baseline<br /> <br /> ===VAD===<br /> <br /> * Code ready, migrate to the VAD code framework<br /> <br /> ===Scoring===<br /> * g-score based on MLP is done<br /> * t-score based on linear regression improves the performance<br /> <br /> ==Word to Vector==<br /> <br /> * Dimension of low space varies from 10-100<br /> * 8-thread word vector generation is 3 times faster than the LDA.<br /> <br /> ==LM development==<br /> <br /> ===NN LM===<br /> <br /> * Character-based NNLM (6700 chars, 7gram), 500M data training done.<br /> :* Non-boundary char LM is better than boundary char LM<br /> <br /> * Investigate MS RNN LM training<br /> <br /> <br /> ==QA==<br /> <br /> ===FST-based matching===<br /> :* Word-based FST 1-2 seconds with 1600 patterns. Huilan's implementation &lt;1 second. ????<br /> :* Char-FST Implementation is done. Not so effective.<br /> <br /> <br /> ===Speech QA===<br /> * Investigate determinization of G embedding</div>

Cslt