ProSign inJoin
Firmware

sp1-usb-audio

Hype27/100

USB audio out for the Teenage Engineering SP-1: a complete Zephyr firmware, the UAC2 glue and packet regulator, and an optional low-CPU driver for the nRF52840

sp1-usb-audio is custom firmware for the Teenage Engineering / KANO Stem Player (SP-1), by @ryanmgilmore. Open source under Apache-2.0, 5 stars on GitHub.

forTeenage Engineering / KANO Stem Player (SP-1)·by@ryanmgilmore

Unclaimed. Is this yours? Sign in with GitHub as @ryanmgilmore to claim it.

Downloads
0
Stars
5
Fork
1
Commits
2
Last commit
Get the source

No release files yet. Build from source.

Apache-2.0· ryanmgilmore/sp1-usb-audio·downloads come from the creator’s GitHub releases

2
Commits
0
Releases
2 commits and 0 releases in the last year
Activity · last 12 months

sp1-usb-audio

USB audio out for the Teenage Engineering SP-1: 48 kHz, 16-bit stereo from the SP-1 to a computer, as a class-compliant USB audio input. No driver on the host.

This is for people writing their own firmware for the SP-1 on Zephyr. It explains how USB audio out works, gives you the code that does it, and records what went wrong along the way so you don't have to find it again. The code comes from an SP-1 firmware that has shipped USB audio in its default build since September 2026.

Start here

sp1/ is a complete SP-1 firmware: it plays a 1 kHz tone through the speaker, at the master volume on Vol+/Vol-, and to the computer over USB at a fixed level. It has everything an SP-1 firmware needs besides the audio: power on and off, the watchdog, the charge gauge. Build it, flash it, and record the tone on your computer — then copy the parts you need into your own firmware. docs/ON-THE-SP-1.md walks through it.

The short version

  • Stock Zephyr already does USB audio. Its UAC2 class (zephyr,uac2) on the nRF52840's stock USB driver is all this needs. Nothing here has to be patched to get sound out.
  • What you do have to write is the part that sizes packets: 47, 48 or 49 frames per 1 ms, to follow the drift between the audio clock and the host's. docs/HOW-IT-WORKS.md.
  • The patched driver is optional. It cuts the USB stack's CPU cost from ~18% to ~0, which matters only for firmware short of CPU. Which to use, and what each costs: docs/STOCK-OR-FAST-PATH.md.

What is here

sp1/the SP-1 firmware: tone to speaker and USB, with power, watchdog and LEDs
usb-audio/the USB audio code itself, shared by both firmwares: the glue and the packet regulator
module/nrf-usbd-isofast/the optional fast-path driver for Zephyr's nRF USB driver
docs/ON-THE-SP-1.mda walkthrough of the SP-1 firmware: clocking, the USB node, where the calls go
docs/STOCK-OR-FAST-PATH.mdwhich driver to use, measured, with the trade-offs of each
docs/HOW-IT-WORKS.mdthe pieces and where each comes from
docs/LESSONS.mdwhat went wrong on the way, and the fixes
test/host tests for the regulator (any C compiler)
extras/nrf52840dk/the same USB side on a Nordic dev board, with no SP-1 code around it — the smallest version to read

Where it has been tested

  • On an SP-1, with macOS: the regulator, the glue and the fast-path driver, as part of a full firmware. Streams clean under full synth load; stops cleanly when the cable is pulled.
  • Not tried: Windows, Linux, and resuming from a host that suspends without closing the stream.
  • The SP-1 firmware in sp1/ has run on an SP-1 with both the stock driver and the fast path: power, tone, volume, headphones, a clean USB recording and power-off all checked on each. Not yet tried there: unplugging mid-stream.
  • The dev-board version has been built with both drivers, not run.

Licence

Apache-2.0 (LICENSE), as Zephyr. The driver in module/nrf-usbd-isofast/ is Nordic Semiconductor's, modified; see NOTICE.