I had a look at recognize, but i couldn't seem to get it to output actually text, it just seems to display phonetics. Any ideas on how to input a sound input from a microphone to output text from it?
This external is for both words and phonemes.
The easiest and most efficient way is to specify your words and/or their order in a grammar text file. Look at any *.gram file example and do the same thing with your own words. These files are in the distribution.
Basically each word has to be separated by a parenthesis; each possibility (or) by a I.
For advanced users, more info about the grammar syntax here :
You first have to record the voice into a buffer~ using record~ for instance. You can then do the speech recognition. Try using two instances and silence detection for switching from one to another if you want an audio stream to be recognized.