ryanontheinside

ryanontheinside on Hacker News

Re:Show HN: Demon – open-source real-time music diffusion engine, 25Hz... on Wednesday May 27, 2026 @03:23PM (#48299201)

oh and as for time - this is several months of work, but is the culmination of several years of audio adjacent genai work

Re:Show HN: Demon – open-source real-time music diffusion engine, 25Hz... on Wednesday May 27, 2026 @03:22PM (#48299188)

lets talk philosophy, bro

Re:Show HN: Demon – open-source real-time music diffusion engine, 25Hz... on Wednesday May 27, 2026 @03:21PM (#48299181)

biggest unexpected unlock i would have to say is the shared mutable curves that bypass the ringbuffer drain. i wouldnt have thought of that without the throughput that preceded it

Re:Show HN: Demon – open-source real-time music diffusion engine, 25Hz... on Wednesday May 27, 2026 @03:20PM (#48299150)

thank you kindly

Re:Show HN: Demon – open-source real-time music diffusion engine, 25Hz... on Wednesday May 27, 2026 @03:19PM (#48299140)

Good question. First and foremost, Magenta is autoregressive, and this is not. Practical differences: THIS: - has more controls; output can be highly custom - has very fast response time for updating controls; this approaches real time - runs on consumer cards - is optimized for loops, and full songs, not continuous infinite music

MAGENTA: - can generate music forever, it plainly wins here - has fewer controls - has high parameter update latency (chunks of 2 seconds) - I did get magenta to fit on a 5090, but just barely

Show HN: Demon – open-source real-time music diffusion engine, 25Hz local GPU

Posted by ryanontheinside on Wednesday May 27, 2026 @12:17PM
from the show-hn dept.

https://daydreamlive.github.io/DEMON/

YO,I’m Ryan, lead author. I’ve been contributing open source generative audio stuff for a while now, audio reactive Comfy nodes, extended ACEstep support in Comfy, etc.. I just opened-sourced a new audio project that I've been working on for a few months and I want to tell y'all about it.WHAT IS IS DEMON: Diffusion Engine for Musical Orchestrated NoiseThis is StreamDiffusion but with audio...

( Read More... | 13 comments | 11 points | on HN )

Re:Show HN: Two-week creative lab for developers building with real-ti... on Wednesday February 04, 2026 @10:13AM (#46886805)

these autoregressive models are wild