“ASR Status Report 2016-12-26”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（5位用户的19个中间修订版本未显示）
第10行： 第10行：
 
|Jingyi Lin
 
|Jingyi Lin
 
||  
 
||  
−
*  
+
* Learn and make Dr.Wang's personal web page.
 +
* Prepare for the CSLT's Annual Meeting.
 
||  
 
||  
−
*  
+
* Finish Dr.Wang's personal web page.
 +
* Take photos for menmbers in CSLT.
 
|-
 
|-
  
第19行： 第21行：
 
|Yanqing Wang
 
|Yanqing Wang
 
||  
 
||  
−
*  
+
* implement the detection mechanism by socket
 +
* find best parameters to avoid over-fitting
 +
* add two-class-SVM to the program
 +
* make GUI more pretty and easy to use
 +
* improve the program's robustness
 +
* screenshot:
 +
** [[媒体文件:dataSender.png|data sender]]
 +
**[[媒体文件:dataAnalyser.png|data analyzer]]
 
||  
 
||  
−
*  
+
* write a document on the program
 
|-
 
|-
  
第30行： 第39行：
 
|Hang Luo
 
|Hang Luo
 
||  
 
||  
−
*  
+
* Run joint training and write systemic script and documents
 
||  
 
||  
−
*  
+
* Finish joint training documents
 +
* Conclude joint training experiments result
 +
* Make a review on mixlingual
 
|-
 
|-
  
第39行： 第50行：
 
|Ying Shi   
 
|Ying Shi   
 
||  
 
||  
−
*  
+
* crawl corpus from internet.(I don't know whether the corpus is right or not.......)
 +
* make new LM(complete)
 +
* train new AM(complete)
 +
* a part of TRP
 
||  
 
||  
−
*  
+
* finish the TRP
 
|-
 
|-
  
第49行： 第63行：
 
|Yixiang Chen   
 
|Yixiang Chen   
 
||  
 
||  
−
*  
+
* Prepare the input of speech data (trick of block segmentation)
 +
* Complete the init version on max-margin SRE.
 +
* Write TRP-20160012 "基于Kaldi i-vector的说话人识别系统使用说明"'''[delivery]'''.
 
||  
 
||  
−
*  
+
* Prepare the thesis proposal.
 +
* Integrate CNN + max-margin.
 
|-
 
|-
  
第58行： 第75行：
 
|Lantian Li   
 
|Lantian Li   
 
||  
 
||  
−
*  
+
* Deep speaker embedding
 +
** Prepare two datasets and make the i-vector baselines.
 +
* Write TRP-20160012 "基于Kaldi i-vector的说话人识别系统使用说明"'''[delivery]'''.
 +
* Write book of robustness SRE.
 +
* Wechat open account.
 
||  
 
||  
−
*  
+
* Deep speaker embedding.
 +
* Write book.
 +
* Replay detection on INTERSPEECH chanllenge.
 
|-
 
|-
  
第67行： 第90行：
 
|Zhiyuan Tang  
 
|Zhiyuan Tang  
 
||  
 
||  
−
*  
+
* TRP of "How to Config Kaldi nnet3 (in Chinese)", not finished yet;
 +
* outline of TRP for "Multi-task Recurrent Model for True Multilingual Speech Recognition";
 +
* Generative models, part of Chapter Deep Learning.
 
||  
 
||  
−
*  
+
* Finish the above 3 writings.
 
|-
 
|-
  
 
|}
 
|}
 +
 +
  
  
 
-------------------------------
 
-------------------------------
 +
 +
 +
 +
 +
{| class="wikitable"
 +
!Date!!People !! Last Week !! This Week
 +
|-
 +
| rowspan="7"|2016.12.19
 +
 +
 +
 +
|Jingyi Lin
 +
||
 +
* --
 +
||
 +
* Concentrate on checking the cslt.book.
 +
* Prepare for the annual convention.
 +
|-
 +
 +
 +
|-
 +
|Yanqing Wang
 +
||
 +
* build a data sender ( read & generate txt files of distracted feature )
 +
* build a data analyzer ( detect the modification of files and make response ( show tokens ) )
 +
* screenshot:
 +
**[[媒体文件:GUI_distracted.png|distraction]]
 +
**[[媒体文件:GUI_focus.png|focus]]
 +
||
 +
* (maybe) replace the detection mechanism by socket
 +
* find best parameters to avoid over-fitting
 +
* add two-class SVM to the program
 +
* make GUI more pretty and easy to use
 +
|-
 +
 +
 +
 +
 +
|-
 +
|Hang Luo
 +
||
 +
* Compare decode result between mono and bi LM, and the decode result ues bi LM before and after joint
 +
* Choose wrong decode sentence and find its difference between baseline and shareGMM baseline
 +
* Finished ML book
 +
||
 +
* Continue joint training analysis work, but I'm very confused about how to improve
 +
|-
 +
 +
 +
|-
 +
|Ying Shi 
 +
||
 +
* some work about kazak lm
 +
* crawl data from kazak internet
 +
||
 +
* run new AM by current speech data
 +
* get more corpus from internet
 +
* use current corpus make LM and decode
 +
|-
 +
 +
 +
 +
|-
 +
|Yixiang Chen 
 +
||
 +
* Leanring tensorflow
 +
* coding pair wise net use tensorflow
 +
* alter CNN
 +
||
 +
* coding CNN connect pair wise
 +
* Dealing with the issue of different lengths of voice
 +
|-
 +
 +
 +
|-
 +
|Lantian Li 
 +
||
 +
* LRE challenge on AP16-OL7.
 +
* Jeju for APSIPA16.
 +
||
 +
* LRE on AP16-OL7.
 +
* Deep speaker embedding.
 +
|-
 +
 +
 +
|-
 +
|Zhiyuan Tang
 +
||
 +
* Jeju for APSIPA16.
 +
||
 +
* A speech about recent ASR improvements.
 +
* A supplementary TRP for "Multi-task Recurrent Model for True Multilingual Speech Recognition".
 +
|-
 +
 +
|}

2017年1月4日 (三) 07:28的最后版本


Date People Last Week This Week
2016.12.26


Jingyi Lin
  • Learn and make Dr.Wang's personal web page.
  • Prepare for the CSLT's Annual Meeting.
  • Finish Dr.Wang's personal web page.
  • Take photos for menmbers in CSLT.
Yanqing Wang
  • implement the detection mechanism by socket
  • find best parameters to avoid over-fitting
  • add two-class-SVM to the program
  • make GUI more pretty and easy to use
  • improve the program's robustness
  • screenshot:
  • write a document on the program
Hang Luo
  • Run joint training and write systemic script and documents
  • Finish joint training documents
  • Conclude joint training experiments result
  • Make a review on mixlingual
Ying Shi
  • crawl corpus from internet.(I don't know whether the corpus is right or not.......)
  • make new LM(complete)
  • train new AM(complete)
  • a part of TRP
  • finish the TRP
Yixiang Chen
  • Prepare the input of speech data (trick of block segmentation)
  • Complete the init version on max-margin SRE.
  • Write TRP-20160012 "基于Kaldi i-vector的说话人识别系统使用说明"[delivery].
  • Prepare the thesis proposal.
  • Integrate CNN + max-margin.
Lantian Li
  • Deep speaker embedding
    • Prepare two datasets and make the i-vector baselines.
  • Write TRP-20160012 "基于Kaldi i-vector的说话人识别系统使用说明"[delivery].
  • Write book of robustness SRE.
  • Wechat open account.
  • Deep speaker embedding.
  • Write book.
  • Replay detection on INTERSPEECH chanllenge.
Zhiyuan Tang
  • TRP of "How to Config Kaldi nnet3 (in Chinese)", not finished yet;
  • outline of TRP for "Multi-task Recurrent Model for True Multilingual Speech Recognition";
  • Generative models, part of Chapter Deep Learning.
  • Finish the above 3 writings.





Date People Last Week This Week
2016.12.19


Jingyi Lin
  • --
  • Concentrate on checking the cslt.book.
  • Prepare for the annual convention.
Yanqing Wang
  • build a data sender ( read & generate txt files of distracted feature )
  • build a data analyzer ( detect the modification of files and make response ( show tokens ) )
  • screenshot:
  • (maybe) replace the detection mechanism by socket
  • find best parameters to avoid over-fitting
  • add two-class SVM to the program
  • make GUI more pretty and easy to use
Hang Luo
  • Compare decode result between mono and bi LM, and the decode result ues bi LM before and after joint
  • Choose wrong decode sentence and find its difference between baseline and shareGMM baseline
  • Finished ML book
  • Continue joint training analysis work, but I'm very confused about how to improve
Ying Shi
  • some work about kazak lm
  • crawl data from kazak internet
  • run new AM by current speech data
  • get more corpus from internet
  • use current corpus make LM and decode
Yixiang Chen
  • Leanring tensorflow
  • coding pair wise net use tensorflow
  • alter CNN
  • coding CNN connect pair wise
  • Dealing with the issue of different lengths of voice
Lantian Li
  • LRE challenge on AP16-OL7.
  • Jeju for APSIPA16.
  • LRE on AP16-OL7.
  • Deep speaker embedding.
Zhiyuan Tang
  • Jeju for APSIPA16.
  • A speech about recent ASR improvements.
  • A supplementary TRP for "Multi-task Recurrent Model for True Multilingual Speech Recognition".
取自“http://index.cslt.org/mediawiki/index.php?title=ASR_Status_Report_2016-12-26&oldid=25379”