Add mfcc function

[jinsu35]

Hi, is this implemented yet? also I'm wondering if chroma, contrast, and tonnetz will be available soon. thanks!

[dhrebeniuk]

@jinsu35 , currently istft in progress. After I'm take in work this function.

[rahul140490]

Hi, including mfcc feature extraction in this library would be a great help, but I was wondering if first and second derivative of mfcc data can also be included. Again, thanks a ton for your work.

[dhrebeniuk]

@rahul140490 , I finished istft, so I'm planning mfcc for next release.

[rahul140490]

Good to know. But may I know when you're planning the next release? And would you also consider first and second derivatives?

[dhrebeniuk]

@rahul140490 , I little bit saw code in librosa, it's used scipy. Thinking how much better use that code.

[rahul140490]

There is this project - https://github.com/nahuelproietto/phoneme-recognition-ios/tree/master/BeatVox, which also has mfcc extraction but not second derivative. May be helpful to you.

[dhrebeniuk]

@rahul140490 , maybe, but main goal of this project, to be same as librosa in python.

[rahul140490]

Yes, I agree. The first and second derivatives of mfcc that I mentioned are part of LibRosa too. Like this - mfcc = librosa.feature.mfcc(y=y, sr=sr) mfcc_delta = librosa.feature.delta(mfcc, order=1) mfcc_delta2 = librosa.feature.delta(mfcc, order=2)

The delta and delta2 are the first and second derivatives. So I was requesting you about this functionality if you can please include them too.

[rahul140490]

Hi, is it implemented yet?

[dhrebeniuk]

@rahul140490 , hello @rahul140490 , not implemented yet.

[rahul140490]

I hope it gets implemented soon, kinda in need of something this good and effective for audio preprocessing. Thanks in advance and all the best.

[dhrebeniuk]

@rahul140490 , I'm trying allocate time for this, it's useful thing for mobile.

[dhrebeniuk]

@jinsu35 , @jinsu35 , please try new version 0.0.4, I implemented basic mfcc function, need test it

[rahul140490]

thanks a lot.