I need to create application which uses speech recognition. At first I thought about using CMU Sphinx (PocketSphinx or, possibly, Sphinx4). Later I thought about JSAPI. (Sphinx somehow uses JSAPI but I don't know what is the difference between JSAPI and Sphinx).
PS There are some other things which I'd like to know:
1) How to load some settings from file (I guess nowadays configuration files are created with the use of XML but I dunno)?
2) How to maintain algorithm which is used by MIDlet which involves JSAPI? I mean there are some different things which my MIDlet needs to do. I guess it is good habit to divide different goals into separate parts of code (due to object-oriented programming).
In my case there are some different things:
a) speech recognition of audio input, i.e. changing input audio stream into output text string b) analysis of that text string and according to this string choosing the proper transition in my algorithm
In general I have written my algorithm on sheet of paper and it takes about ten A4 sheets of paper. Because of it I thought there should be some way to write this algoritm maybe outside the code, in some kind of file which would contain this algorithm. Maybe there is other good way to implement this algorithm, not necesarilly in the code.
c) sending of results through httpconnection with the use of POST method
d) receiving in on TomCat on server
3) Which method should I use to receive the recognized speech? I found these:
a) FinalRuleResult, b) Result -> getBestToken, c) getSpokenText, d) ResultToken of RuleGrammar
4) Can you give me any full examples of JSAPI usage? (Not just short parts of code like in this JSAPI guide)?