4 ms·
Doesn't it suck only because it hasn't been well trained on a variety of voices?
Doesn't it suck only because it hasn't been well trained on a variety of voices?
That very well could be true.
Pocketsphinx is pretty far behind the state of the art in speech recognition algorithms. It also doesn't have a very large acoustic model.
You can greatly improve the recognition by using a restricted language model, however.