multichannel input at default feature extractor

[Enrique Tomas]

Hello,
As I am new at the list, a short introduction....I am Enrique Tomás, using wekinator for my PhD on NIMEs design at the Art University of Linz, Austria.
Thanks for this incredible tool! I am using wekinator for two weeks and I got many inspirations for my work! In particular about the difficult moment of defining my gesture to sound mapping.

Now my question:
I would like to use multiple microphones (let´s say eight) and extract their eight independent audio centroids so I can train my neural network with them. Is it possible to configure wekinator´s default feature extractor for multichannel input? I don´t see the way at the GUI.

thank you in advance,

Enrique

[Jeff Albert]

Enrique,

I have used multiple features from the same audio channel, and I think it should work similarly. My "listener" is running in Max 6 and I just check the "custom OSC feature extractor" box on the Features Setup tab, and put the correct number of features. Then just send the features from your "listener" to the correct OSC port. I simply copied all of the OSC parts of the patch from the examples that were in the Wekinator download.

I guess what I am saying is that you may have to build your own feature extractor, but it isn't too difficult.

My $.02,

Jeff

-- 

Jeff Albert

http://about.me/jeffalbert

--
You received this message because you are subscribed to the Google Groups "wekinator-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wekinator-use...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Enrique Tomas]

Yes, I just throw the possibility of including multichannel input in the TODO list.

I am using timbreID by William Brent with quite good results under Pure Data, sending the analysis to Wekinator.
http://williambrent.conflations.com/pages/research.html

thanks
Enrique