-2

I am trying to do genre based classification of the song.

I have extracted the audio features ( MFCC, Pitch, and Rhythm) for a song.

I trained the three model (ANN, SVM, SVM ) by each features. Three model give the three output (genre of the song).

How can I analysis the output of the three model so that I only get the one result?

Someone
  • 444
  • 3
  • 12
  • [Scikit-learns votingclassifier](http://scikit-learn.org/stable/modules/ensemble.html#votingclassifier). – sascha Jul 29 '16 at 17:44

1 Answers1

1

That's somehow a good question. so i answer it.

Try look at the idea of bagging in Random Forest. These ideas are called Ensemble Methods, that combine the result of other classifier to one last class.

For your question, simply you can get most repeated genre from these 3, to one last classifier. For example if ANN,SVM,SVM respectively output genre1,genre1,genre2, then your final result will be genre1

Masoud
  • 1,343
  • 8
  • 25