1

Can I use a speech recognition software as general purpose software and train it to recognize acoustic models from my regional language? Any specific packages available?

Ganesh Birajdar
  • 65

1 Answers1

2

Sure, you can train a model for your language with CMUSphinx

http://cmusphinx.sourceforge.net

For details see the tutorial

http://cmusphinx.sourceforge.net/wiki/tutorialam

CMUSphinx is available in Ubuntu repository but it is recommended to install latest version from github

http://github.com/cmusphinx

Nikolay Shmyrev
  • 523
  • I was unable to install pocketsphinx using tar.gz file available on above link. When I tried ./configure, I got the error "bash: ./configure: No such file or directory". I guess, somehow I have managed to install only some packages as mentioned below, sphinx-voxforge-hmm-en install sphinx-voxforge-lm-en install sphinx3 install sphinxtrain install python-sphinxbase install ...and now I'm not able to 'import python-pocketsphinx' in python shell. Do I have the pocketsphinx installed? Please let me know if there is some front-end GUI for sphinx, I'm new to programming. – Ganesh Birajdar Jan 31 '15 at 06:00
  • You need to remove all the packages you installed from package manager and try again with the tar.gz you downloaded from github. You need to run ./autogen.sh, not configure. You also need to install sphinxbase before pocketsphinx. For more details see the tutorial http://cmusphinx.sourceforge.net/wiki/tutorial, if you have questions feel free to ask on cmusphinx forum. – Nikolay Shmyrev Jan 31 '15 at 10:21