ASR:2015-12-1 - 版本历史

Zxw：/* Speech Processing */

2015-12-07T07:21:02Z

‎Speech Processing

2015-12-07T07:17:54Z

2015-12-07T06:42:38Z

‎conditioning learning

2015-12-07T06:42:13Z

‎Speech Processing

2015-12-07T06:21:02Z

2015-12-07T06:19:39Z

‎End to End

2015-12-07T06:18:45Z

‎Speech Processing

2015-12-07T05:11:03Z

2015-12-07T05:10:47Z

2015-12-03T07:04:16Z

@@ 第40行： / 第40行： @@
 * AutoEncoder + metric learning
 * binary ivector
 ===multi-GPU===
@@ 第52行： / 第63行： @@
 * wrote nnet3 biglm-decoder for sinovoice.
 * train mpe using wsj and aurara4 --Zhiyong,Xuewei
 ===multi-task===
@@ 第62行： / 第74行： @@
 :* no significant performance improvement observed
 :*http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=483
-: test using extreme data
+: get results with extra input of speech rate info --Zhiyuan
 ==Text Processing==

@@ 第49行： / 第49行： @@
 :*http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=472
 * train 7*2048 tdnn using 4000h data --Mengyuan
 * train mpe using wsj and aurara4 --Zhiyong,Xuewei

@@ 第11行： / 第11行： @@
 ==== conditioning learning ====
-* language vector into multiple layers --zhiyuan
+* language vector into multiple layers --Zhiyuan
 :* a Chinese paper
-* speech rate into multiple layers --zhiyuan
+* speech rate into multiple layers --Zhiyuan
 :*verify the code for extra input(s) into DNN

@@ 第10行： / 第10行： @@
 :* CTC/nnet3/Kaldi
-=== conditioning learning ===
+==== conditioning learning ====
 * language vector into multiple layers --zhiyuan
 :* a Chinese paper

@@ 第9行： / 第9行： @@
 :* MPE
 :* CTC/nnet3/Kaldi
 ====Adapative learning rate method====
@@ 第34行： / 第40行： @@
 * AutoEncoder + metric learning
 * binary ivector
 ===multi-GPU===

@@ 第5行： / 第5行： @@
 ==== Environment ====
-==== End to End ====
+==== End-to-End ====
 *monophone ASR --Zhiyuan
 :* MPE

@@ 第1行： / 第1行： @@
 ==Speech Processing ==
 === AM development ===
 ==== Environment ====
-==== RNN AM====
+==== End to End ====
-*train monophone RNN --zhiyuan
+*monophone ASR --Zhiyuan
-:* end to end MPE
+:* MPE
-:* end to end using nnet3
+:* CTC/nnet3/Kaldi
-:* http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=446
-−
 ====Adapative learning rate method====
 * sequence training -Xiangyu
 :* write a technique report
 ==== Mic-Array ====
@@ 第37行： / 第35行： @@
 * binary ivector
-===language vector===
+=== conditioning learning ===
-* write a paper--zhiyuan
+* language vector into multiple layers --zhiyuan
-:*hold
+:* a Chinese paper
-* language vector is added to multi hidden layers--zhiyuan
+* speech rate into multiple layers --zhiyuan
-:* write code done
+:*verify the code for extra input(s) into DNN
-:* check code
-:*http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=zxw&step=view_request&cvssid=480
-* RNN language vector
-:*hold
-* train with extra input of speech rate info
 ===multi-GPU===

@@ 第31行： / 第31行： @@
 * extract real room impulse to generate WSJ reverberation data, and then train RNN-DAE
-===Ivector&Dvector based ASR===
+===Speaker recognition===
-* learning from ivector --Lantian
+* DNN-ivector framework
-:* CNN ivector learning
+* SUSR
-:* DNN ivector learning
+* AutoEncoder + metric learning
 * binary ivector
-* metric learning
-* LDA-vector Transfer Learning

@@ 第73行： / 第73行： @@
 ==Text Processing==
-/*
+===Work===
-====RNN LM====
+====RNN Poem Process====
-*character-lm rnn(hold)
+* Combine addition rhyme.
-*lstm+rnn
+* Investigate new method.
-:* check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)
+====Document Represent====
-*/
+* Code done. Wait some experiments result.
 ====Neural Based Document Classification====
 ====RNN Rank Task====
 ====Graph RNN====
 :* Entity path embeded to entity.
 *(hold)
 ====RNN Word Segment====
 :* Set bound to word segment.
 * (hold)
 ====Recommendation====
 * Reproduce baseline.
 :*LDA matrix dissovle.
 :* LDA (Text classification & Recommendation System) --> AAAI
 ====RNN based QA====
 *Read Source Code.
 *Attention based QA.
 *Coding.
 ===Text Group Intern Project===