wahibhaq / android-speaker-audioanalysis Goto Github PK

Hi,
It's very interesting your thesis project and i would like to test it.
how can I test it?
thanks

Best Regards

Hi @wahibhaq

Can you please explain how can I use the GaussianMixture.java class, in your android project of speaker detection ?

I tried to import your project into Android Studio, and now the extraction of MFCC features and dumping them into a CSV file, works perfectly. But I cannot understand how to proceed to create a Gaussian Mixture Model from those features.

I tried looking up on what should be provided as weights to the GaussianMixture class, and found the following:
StackExchange - Covariance and Weights
StackOverflow - Gaussian Component Class' weight

From what I understand, is that K-Means clustering would work to cluster different speakers separately, when the MFCC features of audio samples of all the speakers are available.
But how will K-Means clustering work, when audio sample is collected for a particular speaker ?

Please guide if I'm proceeding in the right direction, and correct my understanding if needed.
Thanks.