“ASR Status Report 2017-8-14”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
第132行: 第132行:
 
|Lantian Li   
 
|Lantian Li   
 
||  
 
||  
*  
+
* Visualization and quantification for d-vector [http://cslt.riit.tsinghua.edu.cn/mediawiki/images/e/e2/Spk_seg.pdf].
 +
** phone-aware and phone-blind.
 +
** within speaker variation and between speaker variation.
 +
* Lots of trifles.
 
||  
 
||  
*  
+
* Speaker segmentation task.
 
|-
 
|-
  

2017年8月14日 (一) 05:12的版本

Date People Last Week This Week
2017.8.14 Xiaofei Kang
Miao Zhang
Yanqing Wang
  • TRP uploaded.
  • explore the importance of sparseness structure:
    • After pruning, initialize non-zero values randomly, train.
    • train nnet with 177-dimension hidden layer.
    • result
  • continue exploring the values of trained nnet.
Ying Shi
Yixiang Chen
Lantian Li
Zhiyuan Tang




Date People Last Week This Week
2017.8.7 Xiaofei Kang
  • Finish experiments of 12-style speech with ZhangMiao. (Results are shown in ZhangMiao's CVSS)
  • Complete a part of the recording work: collecting six types of sound from 13 people.
  • Finish the recording work left with ZhangMiao
  • Build a new test website with ZhangMiao
Miao Zhang
  • Finish experiments of 12-style speech with Xiaofei. (Results are shown in CVSS)
  • Build a new test website
  • Recording work
  • Improve the website by decreasing salience segments and replenish other styles
Yanqing Wang
  • retrain experiments finished
  • TRP finished
  • structure V.S. value
Ying Shi
  • setup server for m2asr [finished]
  • design crawler program
  • finish the crawler program
  • CodeMap for Tibetan
Yixiang Chen
Lantian Li
  • Visualization and quantification for d-vector [1].
    • phone-aware and phone-blind.
    • within speaker variation and between speaker variation.
  • Lots of trifles.
  • Speaker segmentation task.
Zhiyuan Tang
  • Some functions of the auto-scoring system rewrited.
  • An app demo with Shuai Zhang.
  • Kaldi book writing.