“ASR:2015-04-20”版本间的差异

2015年4月20日 (一) 04:47的版本

Speech Processing

AM development

Environment

• grid-11 often shut down automatically, too slow computation speed.
• add a server(760)

RNN AM

• details at http://liuc.cslt.org/pages/rnnam.html
• tuning parameters on monophone NN
• run using wsj,MPE

Mic-Array

• investigate alpha parameter in time domian and frquency domain
• ALPHA>=0, using data generated by reverber toolkit
• consider theta

Convolutive network

• HOLD
• CNN + DNN feature fusion

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

• HOLD -Zhiyong
• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=261

Speaker ID

• DNN-based sid --Yiye
• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhangzy&step=view_request&cvssid=327

Ivector based ASR

• hold

• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?step=view_request&cvssid=340
• Ivector dimention is smaller, performance is better
• Augument to hidden layer is better than input layer
• train on wsj(testbase dev93+evl92)

Dark knowledge

• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=264 --zhiyong
• trial on logit matching faild --mengyuan
• adaptation for chinglish under investigation-mengyuan
• unsupervised training with wsj contributes to aurora4 model--xiangyu
• test large database with amida--xiangyu

bilingual recognition

• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=359--zhiyuan

Text Processing

tag LM

• similar word extension in FST

• will check the formula using Bayes and experiment
• add similarity weight

RNN LM

• rnn

• test the ppl and code the character-lm

• lstm+rnn

• check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)

W2V based document classification

• result about norm model [1]
• try CNN model

Translation

• v5.0 demo released

• cut the dict and use new segment-tool

Sparse NN in NLP

• test the drop-out model and the performance gets a little improvement, need some result:
• test the order feature ,need some result:
• large dimension result:http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=lr&step=view_request&cvssid=344

• sparse-nn on 1000 dimension(le-6,0.705236) is better than 200 dimension(le-12,0.694678).

online learning

• modified the listNet SGD

relation classifier

• check the CNN code and contact the author of paper