[svantana]

This is interesting, although the results are underwhelming IMO. I actually had the same idea - finding the latent space of instrument sounds and using that for synthesis - a couple of years back. After countless hours of research I managed to turn it into a commercial software instrument called "GalaXynth" [1]. For me at least, it turned out that the "automatic" latent space (i.e. discovered by autoencoding) isn't that interesting musically, therefore I turned to hand-designing the latent space, which was a gargantuan task that I'm not sure I would do over again if I knew how hard it would be. Anyway if anyone is interested in this type of thing you should get in touch!

[1] https://heartofnoise.com/products/galaxynth/

[edraferi]

Looks like this project made similar trade offs:

> Because the WaveNet decoder is computationally expensive, we had to do some clever tricks to make this experience run in real-time on a laptop. Rather than generating sounds on demand, we curated a set of original sounds ahead of time. We then synthesized all of their interpolated z-representations. To smooth out the transitions, we additionally mix the audio in real-time from the nearest sound on the grid. This is a classic case of trading off computation and memory.

[zubat]

Just wanted to say that I admired GalaXynth when it came out. I still have the demo and mean to give it another go sometime. As you're probably well aware the justifications to buy a synth are hard to pin down and if you made a second run at the concept with a focus on fleshing out "traditional" features I bet there would be some breakthroughs.

[svantana]

Thanks, appreciate the feedback! Others are pointing me in the same directions, so I'm working on adding stuff like filters to push it more in the "canonical" direction.

[matchagaucho]

Pretty cool. Thanks for investing time in this and producing the videos.

The synth performer in me is longing for a traditional ADSR-type of interface, such that I can emphasize the attack and delay of a bowed-string instrument, sustain like a reed, and release like a horn.

Perhaps that's achievable with Galaxynth?

[svantana]

Yes, that was the main feedback I got after releasing it and since version 1.1 there is a modulation section where any parameter can be modulated with envelopes and LFOs, including what sound is played. I never got around to making a video about it, though.

[RickS]

This is such a cool idea! I love the UI you came up with for blending sounds. It's so immediately obvious how it works after seeing/hearing the demo vid. Super awesome.

[catshirt]

wow man this looks awesome. and like an insane amount of work. thanks for sharing!

[tchaffee]

No Linux version? :-(

[svantana]

It's on my todo-list! But as I understand it from other devs, the market for commercial audio software on linux is pretty small.

[microcolonel]

You could probably host the windows vst with Wine.