“2020-04-20”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
 
(2位用户的2个中间修订版本未显示)
第32行: 第32行:
 
|Zhiyuan Tang
 
|Zhiyuan Tang
 
||  
 
||  
*  
+
* Denoising speech with different colors of noise.
*  
+
* Speech book: kws chapter.
 
||
 
||
*  
+
* Continue the enhancement, nda.
 
||
 
||
 
*   
 
*   
第44行: 第44行:
 
|Lantian Li
 
|Lantian Li
 
||  
 
||  
*  
+
* Go on study NDA scoring.(Fast convergence)
*
+
 
||
 
||
*  
+
* Verify the consistent on NDA + NL vs. PLDA scoring.
 
||
 
||
 
*   
 
*   
第92行: 第91行:
 
|Jiawen Kang
 
|Jiawen Kang
 
||  
 
||  
*  
+
* Experiments about loss function, NN depth.
 +
* read more papers about meta learning.
 
||
 
||
*  
+
* Try ft-based and metric-based mate learning.
 
||
 
||
 
*   
 
*   

2020年4月20日 (一) 00:03的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Book chapter for language recognition and emotion recognition
  • Design prior + condition approach for speech enhancement
  • Complete the two chapters of the speech book
Yunqi Cai
  • some simulate experiments on VAE-NDA model
  • check ASR book
Zhiyuan Tang
  • Denoising speech with different colors of noise.
  • Speech book: kws chapter.
  • Continue the enhancement, nda.
Lantian Li
  • Go on study NDA scoring.(Fast convergence)
  • Verify the consistent on NDA + NL vs. PLDA scoring.
Ying Shi
  • Write speech book of minority chapter
  • Summary experiment about speech enhancement
  • Finish speech book
  • Make report about speech enhancement
Haoran Sun
Yue Fan
  • Download all the data of cn2
  • Processed 90% of data
  • Make poster and video of Icassp
  • Get 1000-people data
Jiawen Kang
  • Experiments about loss function, NN depth.
  • read more papers about meta learning.
  • Try ft-based and metric-based mate learning.
Ruiqi Liu
  • Meta-learning experiments.
  • ASR engine authorization for Shengyuan company.
  • Continue last week's work.
Sitong Cheng
  • some tests of DNF baseline
  • Use language information to replace pronunciation score
  • prepare a report
Zhixin Liu
Haolin Chen
  • Experiments on BP denoising
  • Finalize all results