“2024-05-06”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（15位用户的23个中间修订版本未显示）
第6行： 第6行：
 
|Dong Wang
 
|Dong Wang
 
||  
 
||  
−
*  
+
* AIGraph Slides Done
 +
* Slides for several presentations
 
||
 
||
 
*
 
*
第17行： 第18行：
 
|Lantian Li
 
|Lantian Li
 
||
 
||
−
*  
+
* GPU status [https://z1et6d3xtb.feishu.cn/wiki/XGcGwRK5viJmpRkjH9AczIhynCh]
 +
* Projects (AED -> MobileNet, TSE -> Data/Exp)
 +
* ASIP-BUPT (NeuralScoring, CohortSS)
 +
* IS rebuttal
 +
* Rabbit05
 
||
 
||
 
*  
 
*  
第39行： 第44行：
 
|Zhenghai You
 
|Zhenghai You
 
||  
 
||  
−
*  
+
* Model adjustment for Huawei project[https://z1et6d3xtb.feishu.cn/wiki/IQXxwGXTIidQSMkqT6Hcdrqdn8b][https://z1et6d3xtb.feishu.cn/docx/SFZBdrHafohmQJx1ti7c2RZwnuf]
 +
* Data collection for Huawei projects
 
||
 
||
 
*  
 
*  
第49行： 第55行：
 
|Junming Yuan
 
|Junming Yuan
 
||  
 
||  
−
*
+
* KWS project code refactoring
 +
* Control FA experiment baseline result[https://z1et6d3xtb.feishu.cn/docx/Ua0cdv3ano0qHoxN8YvcmRsVn9f]
 
||
 
||
 
*
 
*
第60行： 第67行：
 
|Chen Chen
 
|Chen Chen
 
||  
 
||  
−
*  
+
* Prepare for CNVSRC2024
 +
* About 50 hours new data
 +
* Paper reading
 
||
 
||
 
*  
 
*  
第71行： 第80行：
 
|Xiaolou Li
 
|Xiaolou Li
 
||  
 
||  
−
*  
+
* CNVSRC2024 E-Branchformer test
 +
** better on multi but worse on single
 +
* Paper reading
 
||
 
||
 
*  
 
*  
第82行： 第93行：
 
|Zehua Liu
 
|Zehua Liu
 
||  
 
||  
−
*
+
* Crop size Exp
 +
* AKVSR training
 
||
 
||
 
*  
 
*  
第93行： 第105行：
 
|Pengqi Li
 
|Pengqi Li
 
||   
 
||   
−
*  
+
* Extend Experiment on Timit[https://z1et6d3xtb.feishu.cn/docx/U06ZdLfZOoHXXDxgjdRcIHK7nOd]
 +
** Training and analysis: Assume the models has content bias(Unreliable)
 +
** Try to use larger data training to resist this bias.
 +
** Statistics di-phone On Audio-Mnist
 
||
 
||
 
*
 
*
第104行： 第119行：
 
|Wan Lin
 
|Wan Lin
 
||  
 
||  
−
*  
+
* Neural Scoring
 +
** EAASP in Sunine (training)
 +
** Adjust BCE loss
 +
* Revise graduation paper
 
||
 
||
 
*
 
*
第115行： 第133行：
 
|Tianhao Wang
 
|Tianhao Wang
 
||  
 
||  
−
*  
+
* Neural Scoring: use wespeaker, can't reach the performance of sunine [https://z1et6d3xtb.feishu.cn/docx/BywjdkGvNou12sxQ4dAcxYa9noh]
 +
* SpEx+ training ...
 
||
 
||
 
*  
 
*  
第126行： 第145行：
 
|Zhenyu Zhou
 
|Zhenyu Zhou
 
||  
 
||  
−
*
+
*data collection[https://z1et6d3xtb.feishu.cn/docx/ZgkUdxHbOo7MWoxjDw2c7RIHnYc]
 
||
 
||
 
*
 
*
第136行： 第155行：
 
|-
 
|-
 
|Junhui Chen
 
|Junhui Chen
 +
||
 +
* Graduation paper
 +
* Neural Scoring: use large pretrain model(wav2vec2.0) as test encoder.
 +
** seems useless.(Mix trial EER: 9%)
 
||
 
||
 
*  
 
*  
−
||
 
−
*
 
 
||
 
||
 
*   
 
*   
第148行： 第169行：
 
|Jiaying Wang
 
|Jiaying Wang
 
||  
 
||  
−
*  
+
* data collection for huawei project
 +
* continue cohort experiment [https://z1et6d3xtb.feishu.cn/docx/IeIydyjzJozfUWxvmNfcWASBngg]
 +
* reproduced another baseline DPTNet(waiting for test)
 
||
 
||
 
*  
 
*  
第159行： 第182行：
 
|Yu Zhang
 
|Yu Zhang
 
||
 
||
−
*  
+
* Debugging for evalml (some mismatch in training and testing)
 
||
 
||
 
*
 
*
第182行： 第205行：
 
|Yang Wei
 
|Yang Wei
 
||  
 
||  
−
*  
+
* Double check reliability of the baseline model for children MDD challenge
 
||
 
||
 
*
 
*
第192行： 第215行：
 
|Lily
 
|Lily
 
||
 
||
−
*  
+
* Prepare for AI Radiance ceremony
 
||
 
||
 
*
 
*

2024年5月6日 (一) 11:01的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • AIGraph Slides Done
  • Slides for several presentations
Lantian Li
  • GPU status [1]
  • Projects (AED -> MobileNet, TSE -> Data/Exp)
  • ASIP-BUPT (NeuralScoring, CohortSS)
  • IS rebuttal
  • Rabbit05
Ying Shi
Zhenghai You
  • Model adjustment for Huawei project[2][3]
  • Data collection for Huawei projects
Junming Yuan
  • KWS project code refactoring
  • Control FA experiment baseline result[4]
Chen Chen
  • Prepare for CNVSRC2024
  • About 50 hours new data
  • Paper reading
Xiaolou Li
  • CNVSRC2024 E-Branchformer test
    • better on multi but worse on single
  • Paper reading
Zehua Liu
  • Crop size Exp
  • AKVSR training
Pengqi Li
  • Extend Experiment on Timit[5]
    • Training and analysis: Assume the models has content bias(Unreliable)
    • Try to use larger data training to resist this bias.
    • Statistics di-phone On Audio-Mnist
Wan Lin
  • Neural Scoring
    • EAASP in Sunine (training)
    • Adjust BCE loss
  • Revise graduation paper
Tianhao Wang
  • Neural Scoring: use wespeaker, can't reach the performance of sunine [6]
  • SpEx+ training ...
Zhenyu Zhou
  • data collection[7]
Junhui Chen
  • Graduation paper
  • Neural Scoring: use large pretrain model(wav2vec2.0) as test encoder.
    • seems useless.(Mix trial EER: 9%)
Jiaying Wang
  • data collection for huawei project
  • continue cohort experiment [8]
  • reproduced another baseline DPTNet(waiting for test)
Yu Zhang
  • Debugging for evalml (some mismatch in training and testing)
Wenqiang Du
  • add apeed perturb、larg negative data、human voice interference to update kws model(completed)
  • Some tests have not completed
Yang Wei
  • Double check reliability of the baseline model for children MDD challenge
Lily
  • Prepare for AI Radiance ceremony
Turi
  • Data collection
    • 23K so far
    • Downloaded the data and randomly checking some audios for each users
Yue Gu
Qi Qu
取自“http://index.cslt.org/mediawiki/index.php?title=2024-05-06&oldid=41740”