VoxForge
VoxForge
Hi guys,
I am trying to use last german acoustic model available here
http://www.repository.voxforge1.org/downloads/de/Tags/AcousticModels/voxforge-de-r20140216.tgz
and I am trying to use it in cmu sphinx. I have dictionary and dmp and everything seems to work but the results of decode process are always empty. With debug I have seen that the recognition always retrieves <s>.
Could you kindly help me to figure out the problem?
Thank you in advance.
Best regards
You need to provide more information in order to get an answer.
How exactly are you trying to use the model? Did you try it to decode the file? Share the data you are using then and the exact command line.
Add -rawlogdir <dir> option to pcoketsphinx_continuous in order to dump raw data passed to recognizer. Share raw data too.
Hi,
thanks for your reply. There is my configuration file attached to this message.
I'm not using pocketsphinx but sphinx4.
Let me know if I need to specify other details.
Thanks again
Voxforge-de was trained with the latest most accurate sphinxtrain. To use it you need latest sphinx4 (at least 5alpha or subversion trunk), it will not work in previous versions. Latest version does not use config, it will work just by pointing to the acoustic model folder.
Are you serious? so, I have too much work to do because I am still working on the previous release :( ...
Are you referring to this link
http://cmusphinx.sourceforge.net/wiki/tutorialsphinx4
If I have understood at the end of the page it explans how to use xml configuration, am I wrong?
So, I will try to use new release next days.
Thanks again
