Hulan-2014-02-14
来自cslt Wiki
目录
ASR
ASR Kernel development
TTS
- Voice adaptation
- Male Tree/MAP adaptation completed. 16k Tree-adapted voice sounds the best
- Female Tree/MAP adaptation completed. The original voice sounds the best
- Child voice
- Adapt child voice from Casia using 50 sentences
Dialog system
Statistical approach (QA)
- Mengyuan finished edit distance computing
- Xiaoxi finished fuzzy patten match
- Zhaofang almost finished integration
- We Chat platform updated to support the custom QA system
- Xiaoxi is investigating FST search