2014-05-16 - 版本历史

2014年5月16日 (五) 01:42 Cslt

2014-05-16T01:42:10Z

Cslt：以内容“==Resoruce Building== * Maxi onboard * Release management should be started: Zhiyong (+) * Blaster 0.1 & vivian 0.0 system release == Leftover questions== * Asymmetric...”创建新页面

2014-05-16T01:10:42Z

以内容“==Resoruce Building== * Maxi onboard * Release management should be started: Zhiyong (+) * Blaster 0.1 & vivian 0.0 system release == Leftover questions== * Asymmetric...”创建新页面

新页面

==Resoruce Building==
* Maxi onboard
* Release management should be started: Zhiyong (+)
* Blaster 0.1 & vivian 0.0 system release

== Leftover questions==
* Asymmetric window: Great improvement on training set(WER 34% to 24%), however the improvement is lost on test. Overfitting?
* Multi GPU training: Error encountered
* Multilanguage training
* Investigating LOUDS FST.
* CLG embedded decoder plus online compiler.
* DNN-GMM co-training

== AM development ==

=== Sparse DNN ===
* GA-based block sparsity (+++)
:* Found a paper in 2000 with similar ideas.
:* Try to get a student working on high performance computing to do the optimization

===Noise training===
:* With-clean training done. Much better on clean testing
:* Experiments done. Prepare paper.

===GFbank===

* GFBank sinovoice 1400 MPE stream
* GFBank sinovoice 6000 MPE stream

===Multilingual ASR===

* MPE-based training is not very sensitive to data imbalance for English & Chinese
* Data duplication can trade-off the performance of two languages
* Test sharing shemes

===Denoising & Farfield ASR===

* Baseline: close-talk model decode far-field speech: 92.65
* Will investigate DAE model.

===VAD===

* VAD bug fixed???
* Test frame VAD accuracy

===Scoring===
* online scoring done??
* checked into gitlab?

==Word to Vector==
* Paper writing done

==LM development==

===NN LM===

* Character-based NNLM (6700 chars, 7gram), 500M data training done.
:* Inconsistent pattern in WER were found on Tenent test sets
:* probably need to use another test set to do investigation.

* Investigate MS RNN LM training

==QA==

===FST-based matching===
:* Word-based FST 1-2 seconds with 1600 patterns. Huilan's implementation <1 second.
:* THRAX toolkit for grammar to FST

* Investigate determinization of G embedding
:* Refer to Kaldi new code

@@ 第20行： / 第20行： @@
 ===Noise training===
-:* With-clean training done. Much better on clean testing
+:* More with-clean training completed. 2 conditions left
-:* Experiments done. Prepare paper.
 ===GFbank===
-* GFBank sinovoice 1400 MPE stream
+* 8k train
-* GFBank sinovoice 6000 MPE stream
+:* GFBank sinovoice 1400 MPE stream
 ===Multilingual ASR===
-* MPE-based training is not very sensitive to data imbalance for English & Chinese
+* Test sharing scheme:
-* Data duplication can trade-off the performance of two languages
+:* decision tree share, xent improvement obtained, MPE no improvement (Chinese worse a bit, English a bit better).
-* Test sharing shemes
 ===Denoising & Farfield ASR===
@@ 第38行： / 第49行： @@
 *  Baseline:  close-talk model decode far-field speech: 92.65
 *  Will investigate DAE model.
 ===VAD===
-* VAD bug fixed???
+* DNN-based VAD (24.77) shower better performance than energy based VAD (45.73)
-* Test frame VAD accuracy
 ===Scoring===
@@ 第49行： / 第73行： @@
 ==Word to Vector==
-* Paper writing done
+* Paper submitted
 ==LM development==
 ===NN LM===