About Gaussian Mixture Model and Hidden Markov Model

Hi guys!

I have a lot of birds audios and I started a project where i wanted to identify the birds. I read a few articles in the net and I decided to make feature extraction by mfcc and classification by GMM or HMM. I managed to remove the noise from the clips and used a voice activity detector to take just the sounds I wanted and now I have a lot of clips of 1-2 seconds with birds sound.

I took a code from the net for the MFCC extraction. This codes gives me a vector(array) with some values. But now im stuck in this point. I dont really know what i have to do with this values.

I read something about a toolbox called VoiceBox which I downloaded but I dont think I will need all the archives included and dont really know what to do with them.

Can someone help me? I have now 2 folders one called Train and another called Test, in both folders I have about 25 clips of audio, i can add some more if needed. I want to make a GMM model and try to identify any clip I add to mu GUI.

Any ideas?

Thanks!