Has Anyone Successfully Used CMU Sphinx for Programming by Voice Recognition

Question

I want to start doing some of my coding by voice recognition software (maybe 10-20% of the work I do).

I've seen that some people have had success with Dragon Natural Speaking (DNS) software, but I use a Mac, and unfortunately, Dragon only works on Windows.

Has anyone used the Carnegie Melon open source Sphinx http://cmusphinx.sourceforge.net/ for programming?

Are there other options that I could implement on a Mac? I don't mind dropping a little bit of cash to make this a reality. Ideally it would be a system where I could add in my own commands. (Check out the awesome stuff this guy did, with DNS: https://www.youtube.com/watch?v=8SkdfdXWYaI)

Jaquiss

The answer is 'yes', there is even a bit messy project about that on sourceforge. If done properly with speaker adaptation CMUSphinx can be extremely accurate and helpful. But StackOverflow is probably not the best place to discuss such an open-ended question. – Nahshunn

Answer 1

There is a protoype plugin for IDEA written by JetBrains developers. The work was done during one of their hackathons.

Simons

Answer 2

If you are not fixed with Sphinx, I would recommend Kaldi as an adaptable, compatible open-source speech recognizer. With kaldi you can adapt your own grammar and commands and retrain the underlying models. In addition, there is a python-wrapper that makes Kaldis use easy and convenient.

Cense