“2022-02-21”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
（以“{| class="wikitable" !People !! This Week !! Next Week !! Task Tracking (<font color="red">DeadLine</font>) |- |- |Dong Wang || * || * || * |- |- |Yunqi Cai ||...”为内容创建页面）
 
 
（12位用户的16个中间修订版本未显示）
第5行： 第5行：
 
|Dong Wang
 
|Dong Wang
 
||  
 
||  
−
*
+
* Experiment on IB control with conditional model [http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=wangd&step=view_request&cvssid=847], rough conclusions were obtained.
 +
* Refine the AV speaker recognition theoretical part.
 +
* Review for ICME.
 
||
 
||
−
*
+
* Complete ICME review
 +
* VQMIVC reproduction, update with random mask
 +
* Some missing papers treatment: (1) true nonlinear LDA (2) CycleFlow (3) Thermal-visual database
 
||
 
||
 
*   
 
*   
第16行： 第20行：
 
|Yunqi Cai
 
|Yunqi Cai
 
||   
 
||   
−
*
+
*NSFC Application
 +
*Materials inverse design investigation
 
||
 
||
 
*  
 
*  
第27行： 第32行：
 
|Lantian Li
 
|Lantian Li
 
||  
 
||  
−
*  
+
* Push CNCSRC (Data release and SR baseline)
 +
* Submit Tencent AI Lab project
 +
* Submit M2ASR concluding report
 +
* Write ASVSpoof response
 
||
 
||
−
*  
+
* Submit ASVSpoof response
 +
* Finish Draft of C-P Map paper
 
||
 
||
 
*   
 
*   
第38行： 第47行：
 
|Ying Shi
 
|Ying Shi
 
||  
 
||  
−
*
+
* Speech engrave on overlap speech data
 +
* M2ASR final report
 
||
 
||
−
*
+
* Speech engrave on overlap speech data
 
||
 
||
 
*   
 
*   
第49行： 第59行：
 
|Haoran Sun
 
|Haoran Sun
 
||  
 
||  
−
*  
+
* autoVC with cycle loss [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/1/16/Autovc-cyc.pdf pdf] [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/6/66/Pre.rar demo]
 
||
 
||
−
*  
+
* cycle loss after adverserial training
 +
* VQMIVC
 
||
 
||
 
*   
 
*   
第60行： 第71行：
 
|Chen Chen
 
|Chen Chen
 
||  
 
||  
−
*
+
* Review papers about lip-reading & audio-visual speech recognization
 +
* Prepare data & environment for experiments of AV-Hubert
 
||
 
||
−
*
+
* <-- keep doing these tasks
 
||
 
||
 
*   
 
*   
第71行： 第83行：
 
|Pengqi Li
 
|Pengqi Li
 
||   
 
||   
−
*  
+
* collated the visualization methods that have been reproduced
 +
* some scripts for baseline(cncsrc)
 
||
 
||
−
*  
+
* study feature aggregation
 
||
 
||
 
*   
 
*   
第82行： 第95行：
 
|Weida Liang
 
|Weida Liang
 
||   
 
||   
−
*
+
* Never-before-seen test [http://166.111.134.19:7777/liangwd/paper.html]
 +
* 3~6 spk cycle loss models on wav2vec+seq2seq model
 +
* Rewrite paper and focus on cycle loss
 
||
 
||
−
*
+
* Finish paper framework
 +
* Push test on WER scoring
 
||
 
||
 
*   
 
*   
第92行： 第108行：
 
|Zixi Yan
 
|Zixi Yan
 
||   
 
||   
−
*
+
* Multi-language W2V model features were used for ASR experiments and compared with traditional MFCC features
 
||
 
||
−
*
+
* Asr experiments on different layers of multilingual W2V model
 
||
 
||
 
*   
 
*   
第102行： 第118行：
 
|Sirui Li
 
|Sirui Li
 
||   
 
||   
−
*
+
* Make an experiment plan
 +
* Read the HuBERT paper and code
 
||
 
||
−
*
+
* Finish the hubert-U framework
 
||
 
||
 
*   
 
*   
第113行： 第130行：
 
|Haoyu Jiang
 
|Haoyu Jiang
 
||  
 
||  
−
*
+
*Find the baseline for CN-Celeb speaker identification
 
||
 
||
−
*
+
*Train this baseline and find face recognition baseline
 
||
 
||
 
*   
 
*   
第134行： 第151行：
 
|Renmiao Chen
 
|Renmiao Chen
 
||  
 
||  
−
*
+
* Check CKA
 +
* Do experiment for gender
 
||
 
||
−
*
+
* Do experiment for cross-modal PLDA
 
|-
 
|-
  
 
|}
 
|}

2022年2月21日 (一) 11:31的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Experiment on IB control with conditional model [1], rough conclusions were obtained.
  • Refine the AV speaker recognition theoretical part.
  • Review for ICME.
  • Complete ICME review
  • VQMIVC reproduction, update with random mask
  • Some missing papers treatment: (1) true nonlinear LDA (2) CycleFlow (3) Thermal-visual database
Yunqi Cai
  • NSFC Application
  • Materials inverse design investigation
Lantian Li
  • Push CNCSRC (Data release and SR baseline)
  • Submit Tencent AI Lab project
  • Submit M2ASR concluding report
  • Write ASVSpoof response
  • Submit ASVSpoof response
  • Finish Draft of C-P Map paper
Ying Shi
  • Speech engrave on overlap speech data
  • M2ASR final report
  • Speech engrave on overlap speech data
Haoran Sun
  • cycle loss after adverserial training
  • VQMIVC
Chen Chen
  • Review papers about lip-reading & audio-visual speech recognization
  • Prepare data & environment for experiments of AV-Hubert
  • <-- keep doing these tasks
Pengqi Li
  • collated the visualization methods that have been reproduced
  • some scripts for baseline(cncsrc)
  • study feature aggregation
Weida Liang
  • Never-before-seen test [2]
  • 3~6 spk cycle loss models on wav2vec+seq2seq model
  • Rewrite paper and focus on cycle loss
  • Finish paper framework
  • Push test on WER scoring
Zixi Yan
  • Multi-language W2V model features were used for ASR experiments and compared with traditional MFCC features
  • Asr experiments on different layers of multilingual W2V model
Sirui Li
  • Make an experiment plan
  • Read the HuBERT paper and code
  • Finish the hubert-U framework
Haoyu Jiang
  • Find the baseline for CN-Celeb speaker identification
  • Train this baseline and find face recognition baseline
Ruihai Hou
Renmiao Chen
  • Check CKA
  • Do experiment for gender
  • Do experiment for cross-modal PLDA
取自“http://cslt.org/mediawiki/index.php?title=2022-02-21&oldid=37800”