“Lantian Li 2016-01-11”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
 
(相同用户的3个中间修订版本未显示)
第1行: 第1行:
Weekly Summary
+
* Weekly Report
  
In this week, lots of miscellaneous things have effect the research progress [1-4].
+
In this week, lots of miscellaneous things have effect the research progress [1-4 is out of the research plan].
  
1. Help Prof.Zheng complete and submit the 'replay detection' patent;
+
:* Help Prof.Zheng complete and submit the 'replay detection' patent;
  
2. Prepare the 973 project report, and attend the meeting in Tianjin.
+
:* Prepare the 973 project report, and attend the meeting in Tianjin.
  
3. Help Wang Gang on the DNN-based truncated speech detection.
+
:* Help Wang Gang on the DNN-based truncated speech detection.
  
4. Help Prof.Zheng check and submit the two overview paper.
+
:* Help Prof.Zheng check and submit the two overview paper.
  
5. For the max-margin metric learning task, make some additional experiments.
+
:* For the max-margin metric learning task, make some additional experiments.(~)
  
And I find that there is something wrong on the the NIST-SRE05 label (wav.scp, spk2utt, utt2spk),  
+
And I find that there is something wrong on the the NIST-SRE05 label (wav.scp, spk2utt, utt2spk),  
  
and the SRE05 has been re-labelled.
+
and the SRE05 has been re-labelled.
  
Next Week
+
* To Do next week
  
1. Go on working on the task 5.
+
:* Go on working on the max-margin metric learning, including the dnn-based training.
 +
 
 +
:* Considering how to do the deep speaker embedding. In emergency !

2016年1月11日 (一) 14:47的最后版本

  • Weekly Report

In this week, lots of miscellaneous things have effect the research progress [1-4 is out of the research plan].

  • Help Prof.Zheng complete and submit the 'replay detection' patent;
  • Prepare the 973 project report, and attend the meeting in Tianjin.
  • Help Wang Gang on the DNN-based truncated speech detection.
  • Help Prof.Zheng check and submit the two overview paper.
  • For the max-margin metric learning task, make some additional experiments.(~)
And I find that there is something wrong on the the NIST-SRE05 label (wav.scp, spk2utt, utt2spk),
and the SRE05 has been re-labelled.
  • To Do next week
  • Go on working on the max-margin metric learning, including the dnn-based training.
  • Considering how to do the deep speaker embedding. In emergency !