“Asr-nsfc-weekly-2017-01-16”版本间的差异

• multilingual chars and acoustics processing toolkit has been upgraded with a manual and test recipes.

• work on lexicon segmentation tool.

• 完成了词缀的分割

• 完成哈萨克语识别的剩余工作

• a python toolkit for multilingual chars and acoustics has been esteblished as a version 0.1

• keep revising the charToolkit, and make a multilingual lexicon processing toolkit structure

• 完成汉藏口语句子翻译500句左右。

• 蒙古语发音词典建议，汉藏口语句子翻译。

• 哈萨克语TRP完成，生成新的test集 新test集在语音模型上的ppl为192.5较为合理

• 等新大录好新的test集以后测试

• syllable based spelling corrections are finished for 4k and 200k corpora.
• Kazak acoustic dictionary program is finished.

• keep work on Kazak ASR corpora problems.

藏语拉萨话口语翻译300句。 校对藏语文本语料90000句，蒙语文本语料60000句。

藏语拉萨话口语翻译。 文本正则化处理。