“2021-10-25”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
 
(8位用户的11个中间修订版本未显示)
第16行: 第16行:
 
|Yunqi Cai
 
|Yunqi Cai
 
||   
 
||   
*  
+
* THS2021 experiments
 
||
 
||
 
*  
 
*  
第56行: 第56行:
 
*   
 
*   
 
|-
 
|-
 
  
 
|-
 
|-
|Tiankai Zhi
+
|Chen Chen
 
||  
 
||  
 
*  
 
*  
||  
+
||
 
*  
 
*  
 
||
 
||
第69行: 第68行:
  
 
|-
 
|-
|Chen Chen
+
|Pengqi Li
||  
+
||
 
*  
 
*  
 
||
 
||
*  
+
* audio important maps based mask
 
||
 
||
 
*   
 
*   
第79行: 第78行:
  
 
|-
 
|-
|Jingxin Shen
+
|Qingyang Zhu
 
||   
 
||   
 
*  
 
*  
第87行: 第86行:
 
*   
 
*   
 
|-
 
|-
 +
  
 
|-
 
|-
|Pengqi Li
+
|Weida Liang
 
||   
 
||   
*  
+
* Publish the baseline test data on the website
 +
* Realize the cycle-loss model
 
||
 
||
* audio important maps based mask
+
* More models with different layers and parameters need to be tested
 +
* Test whether the current model is better than baseline
 
||
 
||
 
*   
 
*   
第99行: 第101行:
  
 
|-
 
|-
|Qingyang Zhu
+
|Zixi Yan
 
||   
 
||   
*  
+
* Tibetan text data download
 +
* Hyper-parameters finetuning
 
||
 
||
*  
+
* analyse reason of the problem of wav2vec-u in Tibetan
 +
* Hyper-parameters finetuning
 
||
 
||
 
*   
 
*   
第109行: 第113行:
  
 
|-
 
|-
|Weida Liang
+
|Sirui Li
 
||   
 
||   
 +
* Tibetan language training in wav2vec-u
 +
* Rebuild wav2vec-u experimental environment
 +
||
 +
* Solve the problem of wav2vec-u in Tibetan
 +
||
 +
 +
|-
 +
 +
 +
|-
 +
|Haoyu Jiang
 +
||
 +
* VGGish model test
 +
||
 +
* Voxceleb2 data set collation
 +
* Fine-tuning VGGish model
 +
||
 +
 +
|-
 +
 +
 +
|-
 +
|Ruihai Hou
 +
||
 
*  
 
*  
 
||
 
||
 
*  
 
*  
 +
||
 +
 +
|-
 +
 +
|-
 +
|Renmiao Chen
 +
||
 +
* Download Voxceleb2 video database
 +
* Download Voxceleb1 video database
 +
* Generate data list and divide database
 +
||
 +
* Use some model to get SOTA model
 
||
 
||
 
*   
 
*   

2021年10月31日 (日) 13:57的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Neural PCA paper
  • Neural PCA paper
Yunqi Cai
  • THS2021 experiments
Lantian Li
  • CNC response
  • Project delivery
  • Hard trials paper
Ying Shi
Haoran Sun
Chen Chen
Pengqi Li
  • audio important maps based mask
Qingyang Zhu
Weida Liang
  • Publish the baseline test data on the website
  • Realize the cycle-loss model
  • More models with different layers and parameters need to be tested
  • Test whether the current model is better than baseline
Zixi Yan
  • Tibetan text data download
  • Hyper-parameters finetuning
  • analyse reason of the problem of wav2vec-u in Tibetan
  • Hyper-parameters finetuning
Sirui Li
  • Tibetan language training in wav2vec-u
  • Rebuild wav2vec-u experimental environment
  • Solve the problem of wav2vec-u in Tibetan
Haoyu Jiang
  • VGGish model test
  • Voxceleb2 data set collation
  • Fine-tuning VGGish model
Ruihai Hou
Renmiao Chen
  • Download Voxceleb2 video database
  • Download Voxceleb1 video database
  • Generate data list and divide database
  • Use some model to get SOTA model