“2025-09-29”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（16位用户的21个中间修订版本未显示）
第6行： 第6行：
 
|Dong Wang
 
|Dong Wang
 
||
 
||
−
*  
+
* Travel to Ruian
 +
*  Collecting correction for English version (primary AI)
  
 
||
 
||
第18行： 第19行：
 
|Lantian Li
 
|Lantian Li
 
||
 
||
−
*  
+
* Review AI Book — College Edition (12/53).
 +
* Project matters: HUAWEI SS/AutoBGM; FYT A/V GenreDetect.
 +
* Prepare materials for my professional title evaluation.
 
||
 
||
 
*
 
*
第29行： 第32行：
 
|Ying Shi
 
|Ying Shi
 
||
 
||
−
*  
+
* Thesis
 +
* revise resume
 
||
 
||
 
*  
 
*  
第41行： 第45行：
 
|Zhenghai You
 
|Zhenghai You
 
||
 
||
−
*  
+
* Multi-Step Infer （multi-task search）[https://z1et6d3xtb.feishu.cn/docx/LlgjdMYVbocGD1xjkMSc17Ttnke]
 +
* Huawei 4-mix SS model training
 
||
 
||
 
*
 
*
第51行： 第56行：
 
|Junming Yuan
 
|Junming Yuan
 
||
 
||
−
* Grade 8 AI practice book(done!)
+
* Grade 8 AI practice handbook(done!)
−
* SUPERB benchmark(Speech processing Universal Performance Benchmark)
+
* Further evaluation of our MT-HuBERT model on other speech downstream tasks.
−
** Source Separation downstream task(SI-SDRi):
+
** Based on SUPERB benchmark(Speech processing Universal Performance Benchmark)
−
*** MT-HuBERT: 10.77, HuBERT-BASE: 9.84
+
*** Firstly focused on Source Separation downstream task(still in training)
 +
*** Intermediate results(SI-SDRi at 50K training steps): MT-HuBERT: 10.77, HuBERT-BASE: 9.84.
 
||
 
||
 
*
 
*
第65行： 第71行：
 
|Xiaolou Li
 
|Xiaolou Li
 
||
 
||
−
*  
+
* AVHuBERT s2s finetune debug
 +
* had a fever for some days
 +
* PhD Application
 
||
 
||
 
*  
 
*  
第87行： 第95行：
 
|Pengqi Li
 
|Pengqi Li
 
||
 
||
−
*
+
* Science popularization activities in Changzhi, Shanxi, and returned to the lab on tomorrow.
 
||
 
||
 
*
 
*
第98行： 第106行：
 
|Wan Lin
 
|Wan Lin
 
||
 
||
−
*  
+
* Writing research proposal
 +
* Some Experiments [https://z1et6d3xtb.feishu.cn/wiki/NJLtw3l1Xi2xQBkUpCbcRic9nOh?from=from_copylink]
 
||
 
||
 
*
 
*
第109行： 第118行：
 
|Tianhao Wang
 
|Tianhao Wang
 
||
 
||
−
*  
+
* writing my paper (method and experiments is almost done)
 +
* job interviews
 
||
 
||
 
*
 
*
第120行： 第130行：
 
|Xiaoxue Luo
 
|Xiaoxue Luo
 
||
 
||
−
*  
+
* read some papers on speech separation of unknown number of speakers
 +
* Apply EDA(encoder-decoder based attractor calculation) method to speech separation
 +
** Environment Configuration(done)
 +
** Familiar with the code and make some adjustments to it(in progress)
 
||
 
||
 
*
 
*
第131行： 第144行：
 
|Junhui Chen
 
|Junhui Chen
 
||
 
||
−
*
+
* LLM:
 +
** 2 different reflection paper reading
 +
** Baseline self-reflection metric collection — code completed, data collection in progress.
 +
** Integrating new reflection methods (e.g. beamsearch-based reflection) into the current pipeline in collaboration with @Zhang Yu — work in progress.
 
||
 
||
 
*
 
*
第142行： 第158行：
 
|Jiaying Wang
 
|Jiaying Wang
 
||
 
||
−
*
+
* construct loudness training data: librimix with different loudness source
 +
* loudness order exp[https://z1et6d3xtb.feishu.cn/docx/TDQ9dMYiHoyZnLx6xJDcLg1ZnF3]:
 +
** chain based structure: 2mix 11.92(150 epoch)
 +
** convtasnet structure: still training, 60epoch achieve 12.29
 +
* ctc order exp:
 +
** only use ctc for order, chain based structure: 10.65
 +
* a speculation: Compared with semantic information (provided by CTC), acoustic-oriented information may be more suitable as a basis for separation.
 +
 
 
||
 
||
−
*
+
* try to use speaker info as order
 
||
 
||
 
*
 
*
第153行： 第176行：
 
|Yu Zhang
 
|Yu Zhang
 
||
 
||
−
*  
+
* LLM:
 +
** Baseline self-reflection metric collection — code completed, data collection in progress.
 +
** Integrating new reflection methods (e.g. beamsearch-based reflection) into the current pipeline in collaboration with @chenjunhui — work in progress.
 +
* AED:
 +
** Added humming test and analysis for Huawei.
 
||
 
||
 
*
 
*
第164行： 第191行：
 
|Wenqiang Du
 
|Wenqiang Du
 
||
 
||
−
*  
+
* Primary AI practice book(over)
 +
* Complete the recording of the international version of  AI courses for Jiaoyubu
 
||
 
||
 
*
 
*
第175行： 第203行：
 
|Yang Wei
 
|Yang Wei
 
||
 
||
−
*  
+
* Train English MD model based on Whisper
 
||
 
||
 
*
 
*
第186行： 第214行：
 
|Yue Gu
 
|Yue Gu
 
||
 
||
−
*  
+
* polish the structure of my thesis and continue find jobs
 
||
 
||
 
*
 
*
第195行： 第223行：
 
|Qi Qu
 
|Qi Qu
 
||
 
||
−
*  
+
* Elderly alarm customization.
 +
* SLAM (simultaneous localization & mapping) experiments.
 
||
 
||
 
*
 
*

2025年9月29日 (一) 11:14的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Travel to Ruian
  • Collecting correction for English version (primary AI)
Lantian Li
  • Review AI Book — College Edition (12/53).
  • Project matters: HUAWEI SS/AutoBGM; FYT A/V GenreDetect.
  • Prepare materials for my professional title evaluation.
Ying Shi
  • Thesis
  • revise resume
Zhenghai You
  • Multi-Step Infer （multi-task search）[1]
  • Huawei 4-mix SS model training
Junming Yuan
  • Grade 8 AI practice handbook(done!)
  • Further evaluation of our MT-HuBERT model on other speech downstream tasks.
    • Based on SUPERB benchmark(Speech processing Universal Performance Benchmark)
      • Firstly focused on Source Separation downstream task(still in training)
      • Intermediate results(SI-SDRi at 50K training steps): MT-HuBERT: 10.77, HuBERT-BASE: 9.84.
Xiaolou Li
  • AVHuBERT s2s finetune debug
  • had a fever for some days
  • PhD Application
Zehua Liu
Pengqi Li
  • Science popularization activities in Changzhi, Shanxi, and returned to the lab on tomorrow.
Wan Lin
  • Writing research proposal
  • Some Experiments [2]
Tianhao Wang
  • writing my paper (method and experiments is almost done)
  • job interviews
Xiaoxue Luo
  • read some papers on speech separation of unknown number of speakers
  • Apply EDA(encoder-decoder based attractor calculation) method to speech separation
    • Environment Configuration(done)
    • Familiar with the code and make some adjustments to it(in progress)
Junhui Chen
  • LLM:
    • 2 different reflection paper reading
    • Baseline self-reflection metric collection — code completed, data collection in progress.
    • Integrating new reflection methods (e.g. beamsearch-based reflection) into the current pipeline in collaboration with @Zhang Yu — work in progress.
Jiaying Wang
  • construct loudness training data: librimix with different loudness source
  • loudness order exp[3]:
    • chain based structure: 2mix 11.92(150 epoch)
    • convtasnet structure: still training, 60epoch achieve 12.29
  • ctc order exp:
    • only use ctc for order, chain based structure: 10.65
  • a speculation: Compared with semantic information (provided by CTC), acoustic-oriented information may be more suitable as a basis for separation.
  • try to use speaker info as order
Yu Zhang
  • LLM:
    • Baseline self-reflection metric collection — code completed, data collection in progress.
    • Integrating new reflection methods (e.g. beamsearch-based reflection) into the current pipeline in collaboration with @chenjunhui — work in progress.
  • AED:
    • Added humming test and analysis for Huawei.
Wenqiang Du
  • Primary AI practice book(over)
  • Complete the recording of the international version of AI courses for Jiaoyubu
Yang Wei
  • Train English MD model based on Whisper
Yue Gu
  • polish the structure of my thesis and continue find jobs
Qi Qu
  • Elderly alarm customization.
  • SLAM (simultaneous localization & mapping) experiments.
取自“http://index.cslt.org/mediawiki/index.php?title=2025-09-29&oldid=43501”