“2020-03-09”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
 
(6位用户的7个中间修订版本未显示)
第6行: 第6行:
 
|Dong Wang
 
|Dong Wang
 
||  
 
||  
*  
+
* Almost finished the simulation test for NL
 +
* Completed proof of several important properties of NL
 
||
 
||
*  
+
* Complete the draft of NL paper
 +
* More investigation on subspace DNF
 
||
 
||
 
*   
 
*   
第17行: 第19行:
 
|Yunqi Cai
 
|Yunqi Cai
 
||  
 
||  
*  
+
* Reading papers about SRE back-end score method and Energy based models
 
||
 
||
*  
+
* Run experiment on EBMs
 
||
 
||
 
*   
 
*   
第28行: 第30行:
 
|Zhiyuan Tang
 
|Zhiyuan Tang
 
||  
 
||  
*  
+
* Basic glow for speech normalization.
 +
* Multi-conditioned glow, on phones first.
 
||
 
||
*  
+
* Conditioned glow.
 
||
 
||
 
*   
 
*   
第39行: 第42行:
 
|Lantian Li
 
|Lantian Li
 
||  
 
||  
*  
+
* Basic result on subspace DNF.
 +
* Study on NL scoring.
 
||
 
||
*  
+
* NL scoring.
 
||
 
||
 
*   
 
*   
第50行: 第54行:
 
|Ying Shi
 
|Ying Shi
 
||  
 
||  
*
+
* Double AutoEncoder baseline
 +
* compute fwSNR
 
||
 
||
*
+
* compare fwSNR between Double AutoEncoder Double Flow  and DAE
 +
* experiment about semi-supervise Double Flow
 +
* verify the denoise result via ASR system
 
||
 
||
 
*   
 
*   
第109行: 第116行:
 
|Ruiqi Liu
 
|Ruiqi Liu
 
||  
 
||  
*  
+
* More experiments for different speakers and scenes.
 +
* Continue training the model.
 
||
 
||
*  
+
* Learn Mate-learning,Domain adaptation.
 
||
 
||
 
*   
 
*   
第138行: 第146行:
 
|-
 
|-
  
 +
 +
|-
 +
|Haolin Chen
 +
||
 +
* Optimize double glow model structure
 +
* Calc metrics
 +
||
 +
* Verify denoising for ASR
 +
||
 +
 +
|-
  
 
|}
 
|}

2020年3月9日 (一) 00:05的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Almost finished the simulation test for NL
  • Completed proof of several important properties of NL
  • Complete the draft of NL paper
  • More investigation on subspace DNF
Yunqi Cai
  • Reading papers about SRE back-end score method and Energy based models
  • Run experiment on EBMs
Zhiyuan Tang
  • Basic glow for speech normalization.
  • Multi-conditioned glow, on phones first.
  • Conditioned glow.
Lantian Li
  • Basic result on subspace DNF.
  • Study on NL scoring.
  • NL scoring.
Ying Shi
  • Double AutoEncoder baseline
  • compute fwSNR
  • compare fwSNR between Double AutoEncoder Double Flow and DAE
  • experiment about semi-supervise Double Flow
  • verify the denoise result via ASR system
Wenqiang Du
  • Prepare materials for graduation
Haoran Sun
Yue Fan
  • Optimize baseline model
  • Supplement glow experiment
  • Test the crawler code
  • Do experiments on dnf
  • Complete CN2‘s data download
Jiawen Kang
  • Investigation on adaptation.
  • Arranging experiment results.
  • Learning mate-learning.
Ruiqi Liu
  • More experiments for different speakers and scenes.
  • Continue training the model.
  • Learn Mate-learning,Domain adaptation.
Sitong Cheng
Zhixin Liu
Haolin Chen
  • Optimize double glow model structure
  • Calc metrics
  • Verify denoising for ASR