“ASR Status Report 2017-6-5”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（2位用户的6个中间修订版本未显示）
第8行： 第8行：
 
|Hui Tang  
 
|Hui Tang  
 
||  
 
||  
−
**created a database for speech overlap
+
*created a database for speech overlap
 
* experiments with speech overlap
 
* experiments with speech overlap
−
{| class="wikitable"
+
[http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=tanghui&step=view_request&cvssid=610 here]
−
|-
+
−
! tdnn_word_error
+
−
! tdnn_frame_error
+
−
|-
+
−
|20.00023% (1000,0)/4990
+
−
|
+
−
|}
+
−
{| class="wikitable"
+
−
|-
+
−
! tdnn_vad_word_error
+
−
! tdnn_vad_frame_error
+
−
|-
+
−
|10.9418837675% (539,7)/4990
+
−
|0.208178584613%
+
−
|}
+
 
||  
 
||  
−
*  
+
* align data before overlap
 +
* test in lstm,tdnn
 
|-
 
|-
  
第34行： 第20行：
 
|Yanqing Wang
 
|Yanqing Wang
 
||  
 
||  
−
*  
+
* [http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=wangyanqing&step=view_request&cvssid=609 sparse dnn with different threshold]
 
||  
 
||  
−
*  
+
* continue the experiment
 
|-
 
|-
  
第58行： 第44行：
 
|Yixiang Chen   
 
|Yixiang Chen   
 
||  
 
||  
−
* Ten languages i-vector L-vector + svm performance
+
* OLR17(Ten lans) i-vector / L-vector + svm baseline
 +
* "cough, laugh, hello" data for speaker identification
 
||  
 
||  
−
* cough laugh hello data speaker recognition
+
* go on the two tasks.
 
|-
 
|-
  
第67行： 第54行：
 
|Lantian Li   
 
|Lantian Li   
 
||  
 
||  
−
*  
+
* End-to-end ASV [http://192.168.0.51:5555/cgi-bin/cvss/cvss_request.pl?account=lilt&step=view_request&cvssid=611]
 +
* Deep speaker feature learning.
 
||  
 
||  
−
*  
+
* Speaker diarization.
 
|-
 
|-
  

2017年6月5日 (一) 04:25的最后版本

Date People Last Week This Week
2017.6.5


Hui Tang
  • created a database for speech overlap
  • experiments with speech overlap

here

  • align data before overlap
  • test in lstm,tdnn
Yanqing Wang
  • continue the experiment
Ying Shi
  • mix-lingual structure (without LID)
    • the performance of thch30 is bad than baseline 24.84
    • the performance of kazak is better than baseline 24.37
    • the performance of uyghur is very bad 109.01
  • result analysis
    • pdf mix
    • the language model of uyghur is too weak
  • work for APSIPA2017
Yixiang Chen
  • OLR17(Ten lans) i-vector / L-vector + svm baseline
  • "cough, laugh, hello" data for speaker identification
  • go on the two tasks.
Lantian Li
  • End-to-end ASV [1]
  • Deep speaker feature learning.
  • Speaker diarization.
Zhiyuan Tang
  • basic recipes for OLR17 and MixASR17
  • improve the recipes and publish


Date People Last Week This Week
2017.5.31


Hui Tang
  • after introducing beamforming, zhiyong lets me learn something about how to sound localization
  • Create a database for overlapping speech
Yanqing Wang
Ying Shi
  • work for Apsipa17
    • thchs baseline
    • kazakh baseline
    • uyghur baseline
    • 3 language merge GMM baseline
  • work for Apsipa17
  • graduation thesis
Yixiang Chen
Lantian Li
  • Submit the Interspeech paper.
  • Deep speaker feature learning with a larger ASR model.
  • Prepare the end-to-end ASV.
  • End-to-end ASV.
Zhiyuan Tang
  • Experiments for OC17 mix-asr challenge,baseline done.
  • Spk feature as LID input.
  • Data preparation for AP17 orl challenge.
  • Two challenges done.
取自“http://index.cslt.org/mediawiki/index.php?title=ASR_Status_Report_2017-6-5&oldid=27352”