search

srvk / eesen

The official repository of the Eesen project
http://arxiv.org/abs/1507.08240
Apache License 2.0
822 stars 343 forks source link

RNN LM decoding for Chinese corpus #147

Open Sundy1219 opened 7 years ago

Sundy1219 commented 7 years ago

Now i don't want to use WFST decoder for Chinese ASR, is it possible to use RNN LM for decoding directly ? do you have any solutions ?Looking forward to your reply.Best wishes !!!

fmetze commented 7 years ago

Yes, the tf_clean branch has all the code, we are cleaning up a few things currently. @ramonsanabria has more details

Sundy1219 commented 7 years ago

Thanks for your reply. But i am still confused to the problem that how to use RNNLM to construct a decoder for Chinese ASR. what is the acoustic modeling units ? Chinese characters ? such as "你“ ”好"……？ can you give more information about it ? Looking forward to your reply.Best wishes !!! @fmetze

fmetze commented 6 years ago

I have not tried an RNNLM with Chinese characters, but with a large enough training corpus, you should be able to directly use chinese characters, not? Isn’t this also what Baidu has done in their chinese deep speech version?

On Sep 24, 2017, at 9:06 PM, Sundy1219 notifications@github.com wrote:

Thanks for your reply. But i am still confused to the problem that how to use RNNLM to construct a decoder for Chinese ASR. what is the acoustic modeling units ? Chinese characters ? such as "你“ ”好"……？ can you give more information about it ? Looking forward to your reply.Best wishes !!! @fmetze https://github.com/fmetze — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/srvk/eesen/issues/147#issuecomment-331753294, or mute the thread https://github.com/notifications/unsubscribe-auth/AEnA8czHxbAYVv4lXJLoF0c4o9MZUbh5ks5slvykgaJpZM4Pe1UM.