“FreeNeb Status Report 2017-06-19”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（3位用户的7个中间修订版本未显示）
第21行： 第21行：
 
|-
 
|-
 
|Zhiyong Zhang ||
 
|Zhiyong Zhang ||
−
* Finished phoneme-based TTS test
+
* Finished the run-pass for the phone-based TTS, but poor performance.
 
||
 
||
−
* To revise the text-linguistic feature extraction to avoid the multi-pronunciation problem
+
* To remove the silence from linguistic feature extraction
 
||
 
||
−
* Test LSTM-base KALDI-TTS
+
* Hold
 
||
 
||
−
* To finish the phoneme-based KALDI-TTS</font>(<font color="red">20170616</font>);
+
* To finish the phoneme-based KALDI-TTS</font>(<font color="red">20170616</font>,<font color="red">20170623</font>);
  
 
|-
 
|-
 
|Yang Wei ||
 
|Yang Wei ||
−
* Learn RNN language model training principle and tool.
+
* Crawl 530M universal corpus from zhihu.com(crawler list: https://docs.google.com/spreadsheets/d/1VNtgnnRiez8HjjhpwDXqBvewQZp-tkivOUVA5qk0rPg/edit#gid=0)
 +
* Read source code of LM rescore
 
||
 
||
 
* Read the source code of RNNLM tool
 
* Read the source code of RNNLM tool
第40行： 第41行：
 
|-
 
|-
 
|Dong Wang||
 
|Dong Wang||
−
* Second phase payment done
+
* Some discussion about the partner ship
−
* Some discussion for Awer
+
 
||
 
||
−
* keep an eye on Tonghuashun delivery
+
* Start the first partnership project
 
||
 
||
  
第63行： 第63行：
 
|Zhenlong Han||
 
|Zhenlong Han||
 
* Traing both uni and bi CTC model on data 1000h.
 
* Traing both uni and bi CTC model on data 1000h.
−
* 1000h dnn model has been trained to complete.
+
* 1000h ctc-uni model has been trained to complete.
−
* Wake up with LM and phone-loop model has been test.
+
−
* Write XenC documentation.
+
 
||
 
||
−
* Adjust weight on phone-loop model.
+
* Modify dnn output pdf. but it does not work, I check the code many times, it's seems like the code is no bugs.
 
||
 
||
 
* Wake up.
 
* Wake up.

2017年6月19日 (一) 02:08的最后版本

This Week:

People Last Week This Week Next Week Task Tracing(DeadLine)
Mengyuan Zhao
  • Tonghuashun
  1. Prepare 3rd period delivery.
  • Embedded ASR
  1. Clean up code, finished.
  • Roobo
  1. Cross-compilation.
  • Embedded ASR
  1. Try to add nnet3.
  • Tonghuashun
  1. confirm general testset (20170512)
  • Embedded ASR
  1. Add nnet3 (20170622)
Zhiyong Zhang
  • Finished the run-pass for the phone-based TTS, but poor performance.
  • To remove the silence from linguistic feature extraction
  • Hold
  • To finish the phoneme-based KALDI-TTS</font>(20170616,20170623);
Yang Wei
  • Read the source code of RNNLM tool
  • Similar-pair on Tonghuashun's LM.
Dong Wang
  • Some discussion about the partner ship
  • Start the first partnership project
Mingrui Ruan
  • about FreeNeb
  • 官网建设的商讨
  • 同花顺的项目节点及推进
  • 与爱维尔合作的协议
  • 拜访一个客户
Zhenlong Han
  • Traing both uni and bi CTC model on data 1000h.
  • 1000h ctc-uni model has been trained to complete.
  • Modify dnn output pdf. but it does not work, I check the code many times, it's seems like the code is no bugs.
  • Wake up.
  • Test various output level of CTC
  1. LSTM layer number (20170515)
  2. Frame skipping (20170515)
  • Roobo
  1. First version of "wake up".
取自“http://index.cslt.org/mediawiki/index.php?title=FreeNeb_Status_Report_2017-06-19&oldid=27487”