ASR:2015-05-25

Speech Processing

AM development

Environment

• grid-15 often does not work
• grid-14 often does not work

RNN AM

• details at http://liuc.cslt.org/pages/rnnam.html
• Test monophone on RNN using dark-knowledge --Chao Liu
• run using wsj,MPE --Chao Liu
• run bi-directon --Chao Liu
• train RNN with dark knowledge transfer on AURORA4 --zhiyuan

• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=383--zhiyuan

Mic-Array

• hold
• Change the prediction from fbank to spectrum features
• investigate alpha parameter in time domian and frquency domain
• ALPHA>=0, using data generated by reverber toolkit
• consider theta
• compute EER with kaldi

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

• deliver to mengyuan

• 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 Lin

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

Ivector&Dvector based ASR

• hold --Tian Lan
• Cluster the speakers to speaker-classes, then using the distance or the posterior-probability as the metric
• Direct using the dark-knowledge strategy to do the ivector training.

• 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

• Ensemble using 100h dataset to construct diffrernt structures -- Mengyuan

• http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=264 --Zhiyong Zhang

• adaptation English and Chinglish

• Try to improve the chinglish performance extremly

• unsupervised training with wsj contributes to aurora4 model --Xiangyu Zeng
• test large database with AMIDA
• test hidden layer knowledge transfer--xuewei

bilingual recognition

• hold

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

language vector

• train DNN with language vector--xuewei

Text Processing

RNN LM

• character-lm rnn(hold)
• lstm+rnn

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

W2V based document classification

• make a technical report about document classification using CNN --yiqiao
• CNN adapt to resolve the low resource problem

Translation

• Test the performance of the similar-pair method in bilingual recognition

Order representation

• modify the objective function
• sup-sampling method to solve the low frequence word
• Sort out vectors and do the experiment on objective function convergence
• test on classification task and prediction task

binary vector

• Finish hamming metric binary vector.
• Try to finish binary vector.
• Do test report.

Stochastic ListNet

• To finish writing first edition of emnlp 2015 long paper

relation classifier

• Tune the best model.
• Train on new wordembedding.
• Do some analysis(length of context, track the pooling.)
• Finish the draft.

plan to do

• combine LDA with neural network