I'm trying to make an Android application with speech recognition, but unfortunately google doesn't support my language (MACEDONIAN) and i'm trying to compare two recording sounds.
I'm using http://code.google.com/p/musicg/ to record and compare speech, and i'm stack on initialization the settings for detecting the speech .Some one can tell me how to rewrite this init function for speech detection it's very important to me.. or some other idea how to do that.
this is the initialization for whistle detection
// settings for detecting a whistle
minFrequency = 600.0f;
maxFrequency = Double.MAX_VALUE;
minIntensity = 100.0f;
maxIntensity = 100000.0f;
minStandardDeviation = 0.1f;
maxStandardDeviation = 1.0f;
highPass = 500;
lowPass = 10000;
minNumZeroCross = 50;
maxNumZeroCross = 200;
numRobust = 10;
First off, all you have to do is save your recorded sound into wav and then it is easy to use fingerprint Class from their API https://code.google.com/p/musicg/source/browse/#git%2Fsrc%2Fcom%2Fmusicg%2Ffingerprint
Here is how I'm doing the comparation, one temp recorded WAV sound with all my wav sound in my data.
My understanding is that the musicg DetectionApi, as it stands, is only meant to analyse a single chunk of sound and tell you whether it contains that type of sound. Such as the included whistle or clap api examples. I.e is it a clap/is it a whistle.
With musicg, the best you could probably do would be recognise whether the sound is a voice or not.. though even this might be beyond the DetectionApi.
Since you said the google api doesn't support Macedonian, perhaps you could try Pocketsphinx, which is mentioned in this stackoverflow article.
and here is how I'm saving my temp recorded sound into the wav format: