search

cmusphinx / sphinx4

Pure Java speech recognition library
cmusphinx.sourceforge.net
Other
1.41k stars 586 forks source link

How to use models given here https://sourceforge.net/projects/cmusphinx/files/Acoustic%20and%20Language%20Models/US%20English/ #58

Closed ivish closed 8 years ago

ivish commented 8 years ago

I am having hard time to integrate models given here. I tried to integrate model cmusphinx-en-us-5.2 in pocketsphinx-5prealpha. Since this model is missing 'sendump' file.. I was getting error on running make.. Can I find some doc where I can understand the meaning of the files inside a model. Thanks

gorinars commented 8 years ago

'sendump' is a compressed version of 'mixture_weights'

Check here for the acoustic model format.

Note that the model folder is not needed on 'make' stage. You compile pocketsphinx as usually and then specify the model folder in '-hmm' option. You may also find this discussion useful.

If all above does not work, please post the command you run and the log file you get.

ivish commented 8 years ago

Thanks :) It helped

gorinars commented 8 years ago

You are welcome. Feel free to ask if there are other questions (and close this issue, as well)