The CMUSphinx toolkit is a venerable speech recognition toolkit with various tools used to build speech applications. We are currently trying to consolidate it to reduce confusion. To that end, these are the currently maintained components:
- Pocketsphinx — lightweight recognizer library written in C
- Sphinxtrain — acoustic model training tools
We recommend that you use the latest development code:
Of course, many things are missing. Things like building a phonetic model capable of handling an infinite vocabulary, postprocessing of the decoding result, sense extraction and other semantic tools should be added one day. Probably you should take it on.
The following resources are the main ones for CMUSphinx developers: