To be fair, it doesn't seem to be mentioning this until the "Purpose-built on Compute Module 5" section which is more than halfway through the entire landing page, I was similarly lost until I got to that section and actually understood what I was looking at :)
Looks really fleshed out and fun though, now I just need to get my hands on a proper carrier board it seems, as the audio out of the box for standard CM5 development I/O board is kind of horrible.
Oh, yeah I actually missed that fully, that it expects a USB-connection. Would be a lot of fun and I could find a lot of uses for something like this in my DAWless setup, but with audio-over-USB there isn't much I could do at the moment, but since it's Linux, just changing the output (once I have a new carrier board...) shouldn't be so hard. Thanks for sharing it :)
I wanted another multi-part hardware synth, so I thought it would be fun to see what I could come up with using the parts I already had, particularly the touchscreen.
Loosely inspired by Norns, Brume is a four-part multi-timbral synthesizer that runs on a Raspberry Pi Compute Module 5. It has four synth engines and six voices per part, a shared filter section feeding an effects chain, a sandboxed Lua scripting layer, and a 10.1-inch touchscreen UI. A single USB cable presents the device to a host computer as a class-compliant audio and MIDI interface ala Elektron Overbridge.
Quick BoM to get started: cm5+carrier, any of the endless 10-inch hdmi touchscreens for RPI, and a midi controller. Novation Launch Control XL 3 and Korg nanoKONTROL2 are supported out-of-the box.
* While you can run this on a Raspberry Pi 5, Pi 5s don't expose USB OTG. The carrier board is what makes the class-compliant Audio+MIDI over USB work.
You may already know this, but the voice engines implementations have a lot of room for performance improvement, and you could probably squeeze out even more voices, routing patterns, secondary features, etc if you tighten them up.
They're written clearly but kind of like textbook/reference implementations rather than the more aggressively fine-tuned versions you'd see in proprietary commercial synths.
You can probably coach Claude through improving them, but you generally need to do that explicitly.
Lowest hanging fruit for your design would be to have it organize control flow (and apply effective math tricks) to minimize branching and to organize data references with cpu cache more in mind.
On a more ambitious scale, working over blocks instead of samples, and using vector instructions where practical would go a long way. The latency impact of working on small blocks (i.e ~8/16/32 samples) instead of individual samples is irrelevant but the performance opportunities that open up for many algorithms is huge.
Exactly the kind of feedback I'm looking for, since this is an educational, hobby-class project for myself. I'm just treating the development of it formally as a product to stay disciplined.
Make no mistake, I'm not promoting this thing like a product, though. I'm aware there's a big gulf between this and a commercialized purpose-built product.
That said, there is a ton of headroom right now on the 8GB CM5 reference platform and I felt some of these optimizations were overkill in this pre-1.0 phase. The FX bus is already block-based, but the voice synthesis path remains pretty simple. Some of the block scaffolding for the better approach exists and I'll plan to evolve in that direction.
Just out of curiosity, and without judgement, how much ai was involved in building this (ie was it vibed)? I'm trying to get a better idea of the kind of stuff people are making.
No hesitation to enthusiastically share that it was actually a parallel goal to see just how deeply I could go while keeping it stable, performant, and extensible for the long term.
I ended up building a lot of ancillary agent infrastructure to push the determinism in all the fanned out work slices, too.
Hey, thanks. Kind of obsessed over that demo for too long.
The Rust engine underneath isn't even Pi-locked. It'll run on macOS/Linux, but the touch UI and USB-audio bridge is device specific so it would be fairly useless.
You might be over-indexing on "build this". It's actually just slapping a few 1-click shopping parts together and running two commands in your shell to have a playable instrument.
I've got a vague idea to maybe pull the FM engine out some day since I love having a DX-style synth that actually has a filter, but my plugin work is generally focused on effects processing and not voicing, so I dunno...
Rest assured, if you click the "READ THE MANUAL" button on the landing page you will see screenshots of all synth engines and configuration screens.
The repo's README also starts with a screenshot.
Thanks.
Looks really fleshed out and fun though, now I just need to get my hands on a proper carrier board it seems, as the audio out of the box for standard CM5 development I/O board is kind of horrible.
This was really designed to be surfaced in a DAW via USB.
Loosely inspired by Norns, Brume is a four-part multi-timbral synthesizer that runs on a Raspberry Pi Compute Module 5. It has four synth engines and six voices per part, a shared filter section feeding an effects chain, a sandboxed Lua scripting layer, and a 10.1-inch touchscreen UI. A single USB cable presents the device to a host computer as a class-compliant audio and MIDI interface ala Elektron Overbridge.
Quick BoM to get started: cm5+carrier, any of the endless 10-inch hdmi touchscreens for RPI, and a midi controller. Novation Launch Control XL 3 and Korg nanoKONTROL2 are supported out-of-the box.
* While you can run this on a Raspberry Pi 5, Pi 5s don't expose USB OTG. The carrier board is what makes the class-compliant Audio+MIDI over USB work.
I hope this can be fun for others!
https://brume.aftertone.co https://github.com/aftertonesignal/brume
You may already know this, but the voice engines implementations have a lot of room for performance improvement, and you could probably squeeze out even more voices, routing patterns, secondary features, etc if you tighten them up.
They're written clearly but kind of like textbook/reference implementations rather than the more aggressively fine-tuned versions you'd see in proprietary commercial synths.
You can probably coach Claude through improving them, but you generally need to do that explicitly.
Lowest hanging fruit for your design would be to have it organize control flow (and apply effective math tricks) to minimize branching and to organize data references with cpu cache more in mind.
On a more ambitious scale, working over blocks instead of samples, and using vector instructions where practical would go a long way. The latency impact of working on small blocks (i.e ~8/16/32 samples) instead of individual samples is irrelevant but the performance opportunities that open up for many algorithms is huge.
Make no mistake, I'm not promoting this thing like a product, though. I'm aware there's a big gulf between this and a commercialized purpose-built product.
That said, there is a ton of headroom right now on the 8GB CM5 reference platform and I felt some of these optimizations were overkill in this pre-1.0 phase. The FX bus is already block-based, but the voice synthesis path remains pretty simple. Some of the block scaffolding for the better approach exists and I'll plan to evolve in that direction.
Thank you.
No hesitation to enthusiastically share that it was actually a parallel goal to see just how deeply I could go while keeping it stable, performant, and extensible for the long term.
I ended up building a lot of ancillary agent infrastructure to push the determinism in all the fanned out work slices, too.
https://thinkwright.github.io/threadmark/
Also: https://brume.aftertone.co/develop.html#agents
The Rust engine underneath isn't even Pi-locked. It'll run on macOS/Linux, but the touch UI and USB-audio bridge is device specific so it would be fairly useless.
You might be over-indexing on "build this". It's actually just slapping a few 1-click shopping parts together and running two commands in your shell to have a playable instrument.
I've got a vague idea to maybe pull the FM engine out some day since I love having a DX-style synth that actually has a filter, but my plugin work is generally focused on effects processing and not voicing, so I dunno...