“2021-12-27”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
 
(7位用户的11个中间修订版本未显示)
第5行: 第5行:
 
|Dong Wang
 
|Dong Wang
 
||  
 
||  
*  
+
* Hard trails paper revised
 +
* Investigate link between AE and flow in the linear case [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/9/91/Linear.png]
 +
* Investigate audio visual pretraining
 
||
 
||
*  
+
* Keep on neural PCA theory
 
||
 
||
 
*   
 
*   
第16行: 第18行:
 
|Yunqi Cai
 
|Yunqi Cai
 
||   
 
||   
*  
+
* AE baseline
 +
* Alignet & AE joint training for Infra to Visible domain transfer
 +
* Gatenet for Visible and Infra latent code fusion [https://qw4sbuvrmj.feishu.cn/docs/doccnNaBy0EAyUTskPQvvN7kgWo]
 
||
 
||
 
*  
 
*  
第52行: 第56行:
 
|Haoran Sun
 
|Haoran Sun
 
||  
 
||  
*  
+
* some contrast experiment
 
||
 
||
*  
+
* design the timbre encoder with a design similar to x-vector
 +
* insert the new encoder into our four-encoder structure
 
||
 
||
 
*   
 
*   
第87行: 第92行:
 
|Weida Liang
 
|Weida Liang
 
||   
 
||   
*  
+
* Finish patent disclosure
 +
* Update the project website and upload never_before_seen result[http://166.111.134.19:7777/liangwd/never_before_seen.html]
 +
* Go on with seq2seq test
 
||
 
||
*  
+
* Finish seq2seq test
 +
* Update paper
 
||
 
||
 
*   
 
*   
第97行: 第105行:
 
|Zixi Yan
 
|Zixi Yan
 
||   
 
||   
*  
+
* Debug bugs encountered when using the fine-tuning model
 
||
 
||
 
*
 
*
第107行: 第115行:
 
|Sirui Li
 
|Sirui Li
 
||   
 
||   
*  
+
* Adjust the parameters to use thchs30 training
 
||
 
||
*  
+
* Learn the wav2vec-u process on Swedish
 
||
 
||
 
*   
 
*   
第118行: 第126行:
 
|Haoyu Jiang
 
|Haoyu Jiang
 
||  
 
||  
*  
+
* Sampling out pictures compute FAR
 
||
 
||
 
*  
 
*  

2021年12月27日 (一) 11:34的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Hard trails paper revised
  • Investigate link between AE and flow in the linear case [1]
  • Investigate audio visual pretraining
  • Keep on neural PCA theory
Yunqi Cai
  • AE baseline
  • Alignet & AE joint training for Infra to Visible domain transfer
  • Gatenet for Visible and Infra latent code fusion [2]
Lantian Li
  • Plan CNCSRC stuffs.
  • Maintain Sunine.
  • Deliver projects.
  • Modify patents.
  • Go on Hard trials paper.
Ying Shi
  • m2asr Kirgiz check
  • continue on speech engrave (text attend speech; HMM based attention)
Haoran Sun
  • some contrast experiment
  • design the timbre encoder with a design similar to x-vector
  • insert the new encoder into our four-encoder structure
Chen Chen
  • Do course project
  • Organize Kirgiz data
  • Do course project
  • Prepare for stochastic process final exam
Pengqi Li
Weida Liang
  • Finish patent disclosure
  • Update the project website and upload never_before_seen result[3]
  • Go on with seq2seq test
  • Finish seq2seq test
  • Update paper
Zixi Yan
  • Debug bugs encountered when using the fine-tuning model
Sirui Li
  • Adjust the parameters to use thchs30 training
  • Learn the wav2vec-u process on Swedish
Haoyu Jiang
  • Sampling out pictures compute FAR
Ruihai Hou
Renmiao Chen
  • Do course project
  • Do course project
  • Prepare for stochastic process final exam