Goran K

Hi, is there a machine learning library available for Max that would perform a function similar to 

 The desired result would be a patcher that slices an audio buffer by transients and categorises the slices by similarity. T-SNE puts them on a 2d map, but that part is purely optional. It would be enough to categorise the slices into an arbitrary number of collections (2-5) based on spectral properties.

Hi, is there a machine learning library available for Max that would perform a function similar to Audio t-SNE?  The desired result would be a patcher that slices an audio buffer by transients and categorises the slices by similarity. T-SNE puts them on a 2d map, but that part is purely optional. It would be enough to categorise the slices into an arbitrary number of collections (2-5) based on spectral properties.

Hi, is there a machine learning library available for Max that would perform a function similar to Audio t-SNE?  The desired result would be

audio-t-sne-in-max

There's no t-SNE algorithm in it, but the FluCoMa project (

 ) has loads of tools for doing exactly what you're talking about. (UMAP is probably the next fanciest algorithm that it does have)

There's no t-SNE algorithm in it, but the FluCoMa project (http://flucoma.org ) has loads of tools for doing exactly what you're talking about. (UMAP is probably the next fanciest algorithm that it does have)

Thanks! I've started checking it out; it looks quite promising.

There's a fairly active forum if you want to pop over for questions and stuff:

The public beta of the second set of tools is going live soon, which has a lot of the bits I think you'd be interested in.

There's a fairly active forum if you want to pop over for questions and stuff:
http://discourse.flucoma.org 

Have you had a look at Catart?
It is included in the Mubu package

Wow, Mubu is great! Right from the start it produced interesting results close to what I'm after. Thanks for that!

Audio t-SNE in Max?