falktx @falktx@mastodon.falktx.com

my music listening experience is sorta ruined now after having to deal with audio de-syncs and clicks a lot in the last few days...

listening to random stuff like
"was that a click? maybe an xrun? does my code need more fixing?"
happens now everytime there is even a small click or noise 😭

now this gets a bit ridiculous, 8 frames for usb audio buffer size.

gets latency quite low, but easily xrun if playing anything complex in FLStudio.

seems stable/sustainable while there is very little stuff running though.
I like the small deviation in those "distance" numbers (it is how far userspace is from the kernel position in the ringbuffer).

Linux as a USB audio interface with minimal latency 😀

Hope users will appreciate the effort, the fact that this is running all inside a Linux embed device is quite something...
No external/dedicated USB Audio chips, it's all on the Linux side done in software.

I will take some big holidays after this 🌅 🏖️

For the curious ones, kernel diff https://github.com/Darkglass-Electronics/radxa-kernel/commit/1c99fb0a7788811a4eb212c3c0bdcb64dc9d955f

and user-space side https://github.com/falkTX/audio-bridge/blob/main/src/audio-device-impl-linux-mmap.cpp

The things we do for a few milliseconds of latency...

Like today, patching the Linux kernel to create a memory-mapped buffer that both the kernel and a JACK client can use at the same time.
(the previous/mainline code was using an alsa virtual device to expose the audio data)

Reduced latency of the "task" by around 20ms ✨

Now just need to deal with ARM memory barriers and other details to ensure a good sync, I think...

Not that anyone cares, but just built and installed Cardinal on the Darkglass Anagram unit for quick testing...

The JTB-Waves patch runs with a stable 58% DSP load.

Makes me think of fancy "build your own FX" possibilities...

This Anagram audio graph is looking complex now that USB audio is in the picture... 😁

Average 35% DSP load with this preset

Look internet, I am on video! 😀

https://www.youtube.com/watch?v=CzaVPM_Ja3U

Very glad and proud to be part of #darkglass team. 🚀

Since I am the only one in the company that has dealt with opensource software licenses before, I am the one doing the page that lists the licenses used.
There's so much stuff hah 😱

How it looks so far: https://www.darkglass.com/pablito-licenses/

Still WIP, but what do you think?
Specially interested on opinions from other opensource developers.
Thanks for the feedback in advance!

Typical audio graph setup on the new Darkglass Anagram.

Some of the FX there are "static" like the globaleq, mixer and looper.

All of it running at 16 frames / 48kHz. Total IO roundtrip latency being 1.333ms (some FX do oversampling and introduce a bit more latency).

This preset runs with around 30% CPU load.

Not shown on graph: USB and Bluetooth audio, coming soon.

After pretty much 1 year of work the result is finally public.

For those that don't know, in early 2024 I joined the Darkglass Electronics company. Bringing the expertise developed through many years with MOD Devices / MOD Audio and my own projects, and started to create together with them a super fancy new Linux-embed based product.

This runs JACK2 inside with mod-host and LV2 plugins. Very soon to start opening up to external plugin developers.
Very exciting! 🎉

https://www.darkglass.com/creation/anagram/

As things with life will now finally start to calm down, its also finally time to try to make some music/noise again.

But I am finding it very messy to use the same system for work, personal development, media consumption etc.

A dedicated system/disk/partition is the way to go, right?

A teaser video of an upcoming product... https://www.youtube.com/watch?v=k18pJAMYSTA

The results of ~1 year of work finally becoming public.
Real announcement later in the month.