“ASR:2015-03-30”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
LM development
Zxw讨论 | 贡献
Speech Processing
 
(1位用户的3个中间修订版本未显示)
第4行: 第4行:
 
==== Environment ====
 
==== Environment ====
 
* grid-11 often shut down automatically, too slow computation speed.
 
* grid-11 often shut down automatically, too slow computation speed.
* GPU has being repired.--Xuewei
+
 
  
 
==== RNN AM====
 
==== RNN AM====
 
* details at http://liuc.cslt.org/pages/rnnam.html
 
* details at http://liuc.cslt.org/pages/rnnam.html
* tuning parameters on monophone NN  
+
* tuning parameters on monophone NN
 +
* run using wsj,MPE 
  
  
 
==== Mic-Array ====
 
==== Mic-Array ====
 
* investigate alpha parameter in time domian and frquency domain  
 
* investigate alpha parameter in time domian and frquency domain  
 +
* ALPHA>=0
  
====Dropout & Maxout & rectifier ====
 
* HOLD
 
* Need to solve the too small learning-rate problem
 
* 20h small scale sparse dnn with rectifier. --Mengyuan
 
* 20h small scale sparse dnn with Maxout/rectifier based on weight-magnitude-pruning. --Mengyuan Zhao
 
  
 
====Convolutive network====
 
====Convolutive network====
 
* HOLD
 
* HOLD
 
:* CNN + DNN feature fusion
 
:* CNN + DNN feature fusion
:* reproduce experiments -- Yiye
 
  
 
====RNN-DAE(Deep based Auto-Encode-RNN)====
 
====RNN-DAE(Deep based Auto-Encode-RNN)====
第29行: 第25行:
 
* http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=261
 
* http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=261
  
====Speech rate training====
 
:* http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=268
 
:* Technical report HOLD.-- Xiangyu Zeng, Shi Yin
 
:* Paper for NCMMSC done
 
 
====Neural network visulization====
 
* http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=324
 
* Technical report done --Mian Wang.
 
  
 
===Speaker ID===   
 
===Speaker ID===   
 
:* DNN-based sid --Yiye
 
:* DNN-based sid --Yiye
 +
:* Decode --Yiye
 
:* http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=327
 
:* http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=327
  
第46行: 第35行:
 
:* Ivector dimention is smaller, performance is better
 
:* Ivector dimention is smaller, performance is better
 
:* Augument to hidden layer is better than input layer
 
:* Augument to hidden layer is better than input layer
:* write paper for interspeech -- Xuewei
+
:* train on wsj(testbase dev93+evl92)
  
 
==Text Processing==
 
==Text Processing==
===LM development===
+
===tag LM===
 
+
====Domain specific LM====
+
====tag LM====
+
* Tag Lm(JT)
+
:* get new script from mx and test 1 tag lm
+
 
* similar word extension in FST
 
* similar word extension in FST
:* experiment done
+
:* check the formula using Bayes and experiment
:* write the paper
+
  
 
====RNN LM====
 
====RNN LM====
 
*rnn
 
*rnn
:* the input and output is word embedding and add some token information like NER..
+
:* code the character-lm using Theano
:* map the word to character and train the lm.
+
 
*lstm+rnn
 
*lstm+rnn
 
:* check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)
 
:* check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)
 
===Word2Vector===
 
  
 
====W2V based doc classification====
 
====W2V based doc classification====
* data prepare.(hold)
+
* corpus ready
====Knowledge vector====
+
* learn some benchmark.
* make a report on Monday
+
  
 
===Translation===
 
===Translation===
 
 
* v5.0 demo released
 
* v5.0 demo released
 
:* cut the dict and use new segment-tool
 
:* cut the dict and use new segment-tool
第82行: 第60行:
 
:* check the code to find the problem .
 
:* check the code to find the problem .
 
:* increase the dimension  
 
:* increase the dimension  
:* use different test set.
+
:* use different test set,but the result is not good.
===QA===
+
  
 
===online learning===
 
===online learning===
 
* data is ready.prepare the ACL paper
 
* data is ready.prepare the ACL paper
 
:* prepare sougouQ data and test it using current online learning method  
 
:* prepare sougouQ data and test it using current online learning method  
====framework====
+
:* baseline is not normal.
* extract the module
+
* composite module
+
* fix the bug
+
====leftover problem====
+
* new inter will install SEMPRE
+

2015年4月1日 (三) 05:39的最后版本

Speech Processing

AM development

Environment

  • grid-11 often shut down automatically, too slow computation speed.


RNN AM


Mic-Array

  • investigate alpha parameter in time domian and frquency domain
  • ALPHA>=0


Convolutive network

  • HOLD
  • CNN + DNN feature fusion

RNN-DAE(Deep based Auto-Encode-RNN)


Speaker ID

Ivector based ASR

Text Processing

tag LM

  • similar word extension in FST
  • check the formula using Bayes and experiment

RNN LM

  • rnn
  • code the character-lm using Theano
  • lstm+rnn
  • check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)

W2V based doc classification

  • corpus ready
  • learn some benchmark.

Translation

  • v5.0 demo released
  • cut the dict and use new segment-tool

Sparse NN in NLP

  • prepare the ACL
  • check the code to find the problem .
  • increase the dimension
  • use different test set,but the result is not good.

online learning

  • data is ready.prepare the ACL paper
  • prepare sougouQ data and test it using current online learning method
  • baseline is not normal.