Live data from Hacker News

The Infinite Drum Machine

experiments.withgoogle.com

21–30 of 46 posts

Re: The Infinite Drum Machine

#23
post #7

Sounds like they copied XLN Audio XO - https://assets.xlnaudio.com/pages/xo/1.2/gui-screens/main_2x... They should at least mention what was the "inspiration". Not cool Google. https://www.xlnaudio.com/products/xo

The idea is hardly novel. http://imtr.ircam.fr/imtr/CataRT is seminal in this regard and much much older.

CataRT has moved here https://forum.ircam.fr/projects/detail/catart-standalone/

Re: The Infinite Drum Machine

#24
post #23
post #7

Earlier quoted context omitted.

The idea is hardly novel. http://imtr.ircam.fr/imtr/CataRT is seminal in this regard and much much older.

CataRT has moved here https://forum.ircam.fr/projects/detail/catart-standalone/

That's just the standalone. CataRT has many incarnations and that is just one.

Re: The Infinite Drum Machine

#26

What do you suppose they mean when they call it "infinite"? I'm not seeing much description of what the concept behind the project is supposed to be.

I think it's supposed to be some kind of latent vector space where the coordinate maps to the sounds' timbre. So in theory, you can move seamlessly between sounds. But I seem to only be able to snap to nearest sound. Edit: they use t-SNE for the embedding so that is why it is discrete. But if they used something like an autoencoder, then you could have a smooth continuum and interpolate between say a dog bark and a billiards clack.

Re: The Infinite Drum Machine

#27
Doesn't seem to have a lot of … drums?

No hi hat, no tom tom. No "taiko" drums which I realize is a bit broad. "Cymbals" nets me three "percussion finger bells" (okay, technically, I suppose?) and one "percussion drums".

Some snare and timpani…

Post reply on HN