Punk Confusion

A split-brain punk card for the Music Thing Modular Workshop Computer.

Switch Up is the word "punk" taken literally: a voltage-controlled Atari Punk Console-inspired synth voice. Switch Middle is Broken Venue, a dirty room and damaged PA treatment for external audio. Switch Down and Pulse In 2 fire short venue-linked vocal calls through the same room engine.

Quick Start

1. Flash uf2/punk_confusion_2mb.uf2 for a standard 2 MB card, or uf2/punk_confusion_16mb.uf2 for a 16 MB card.
2. Patch Audio Out 1 to your mixer. Patch Audio Out 2 as well for stereo room output in Broken Venue mode.
3. For APC mode, set the switch Up and turn Main up. Audio Out 1 is the direct APC voice; Audio Out 2 is the same voice through a fixed CBGB-style room/PA treatment.
4. For Broken Venue, patch audio to Audio In 1, set the switch Middle, set Main, X, and Y near noon, then choose a room with X.
5. Hold the switch Down, or patch gates to Pulse In 2, to inject the current venue's vocal call.

Modes

Switch Up: Atari Punk Console

This is a simple dual-555-style APC model rather than an audio effect.

• Main: APC output volume.
• X: trigger oscillator rate, like APC pot 1. Clockwise is faster.
• Y: monostable one-shot time, like APC pot 2. The useful direction is matched to the tested hardware feel.
• CV In 1: adds to X.
• CV In 2: adds to Y.
• Pulse In 1: hard gate when patched. Unpatched, the APC free-runs.
• Audio Out 1: dry APC output.
• Audio Out 2: APC through a fixed CBGB-style room/PA treatment, with the audience absorption set internally around 33% and a small gain trim.

X clocks the astable trigger oscillator. Y sets the one-shot length. When the one-shot is still high, incoming triggers are ignored, giving the classic skipped, stepped APC behaviour. Main controls the level feeding both APC outputs together.

Switch Middle: Broken Venue

Broken Venue processes external audio from Audio In 1 through a fixed 50/50 dry/wet dirty-room engine.

• Main: input/room gain with soft pickup. Below noon attenuates hot modular signals, noon is about unity, and clockwise boosts quieter headphone or line-level sources. Hot Eurorack-level signals may clip above roughly 2 o'clock.
• X: venue selection in room-length order.
• Y: audience absorption. Clockwise means more bodies in the room, reducing reflected energy and lightly damping the dry side so the 50/50 mix still reads as a room change.
• Audio Out 1: main processed output.
• Audio Out 2: decorrelated stereo room output.

The Y curve is inspired by Rummler/Green/Jurkiewicz/Kahle, "Forget About The Seat Dip Effect" (Forum Acusticum / Euronoise 2025). The firmware uses a cheap three-band approximation: small low-frequency loss, stronger attenuation around 400 Hz-3 kHz, and moderate high-frequency damping.

Venue Order

X selects snapped venue zones:

These are not acoustic models of the real rooms. They are four punk-venue personalities built from one compact delay, reflection, saturation, and filtering engine.

Switch Down: Vocal Calls

Switch Down is a momentary performance layer, not a separate full mode.

• Pressing or holding Down triggers the vocal call for the currently selected venue.
• Releasing Down stops playback, so you can stutter the call manually.
• Pulse In 2 mirrors this behaviour: high gates the call, rising edges retrigger it, and low stops it.
• Driving Pulse In 2 at audio rate can chop the shout into a raw vocal texture. That is intentional.
• Audio In 2 is a beta vocal slice/reverse CV input. When patched, positive voltage selects a later start slice for the next trigger; negative voltage selects a slice and plays it backwards. Unpatched, calls play normally from the start.
• The vocal is routed through the same Broken Venue path and gets an extra send into the room delay so it sits inside the venue.

Audio In 2 is sampled only when the vocal is triggered. It does not scrub an already-playing shout.

The Computer audio inputs clip at about +/-6 V. Hotter control signals, such as a full-range Workshop System Slopes output, are safe but do not give extra slice range: high positive voltages hold the last forward slice, and high negative voltages hold the last reverse slice.

While holding Down, Main edits the saved vocal-call trim rather than the main room input gain. The trim has soft pickup and is multiplied by the saved room gain, so later input-gain changes still scale the shout level. The default trim is midpoint.

Vocal Samples

The embedded calls are original recordings by Adrian Vos, processed for Punk Confusion as 24 kHz mono signed 16-bit PCM and kept around -6 dBFS peak to avoid extra digital clipping after the Colourbox drive.

The source WAVs are kept in samples/, matching the organisation used by other sample-based releases in this repo. The beta firmware includes those calls as factory fallback samples, but the normal user workflow is now WebSerial loading: flash the card once, then change the four vocal calls from a browser without rebuilding or reflashing the UF2.

WebSerial Sample Loader

The WebSerial loader stores uploaded calls in a reserved flash sample bank. Uploaded samples persist across normal restarts and may also persist after reflashing, so the web page includes a simple factory-restore button.

Use it like this:

1. Flash uf2/punk_confusion_2mb.uf2 for a standard 2 MB Workshop Computer card, or uf2/punk_confusion_16mb.uf2 for a 16 MB card.
2. Open web/index.html in Chrome, Edge, or another Chromium-based browser with WebSerial support.
3. Choose the matching card size in the page.
4. Drop one audio file into each venue slot on the page. If you already have a saved punk_confusion_samples.pbank file, drop it into Already have prepared sounds? instead to fill all four slots at once. You can also use Browse a folder of WAVs to choose or drop any folder of .wav files, audition the detected files, and assign four of them to the sample slots.
5. Hold the card switch Down while powering or resetting the card.
6. Wait for confirmation: all LEDs flash three times, then LEDs 1, 3, and 5 stay lit while the card waits for the browser.
7. Press Connect card, choose the Workshop Computer serial device, then press Send sounds to card.
8. Restart the card and use it normally with the new shouts.

During upload the status panel shows four stages: asking whether the card is ready, sending the sounds, waiting while the card saves them, and upload complete. Larger banks can take a little while to write, so wait for the final complete message before restarting the card.

The uploaded samples are stored in flash, so they can persist even after you reflash the firmware. To return to the embedded factory shouts, enter loader mode again and press Use built-in sounds again in the web page. That erases the uploaded sample-bank header, so the firmware falls back to VocalSamples.h.

The WebSerial builds use 24 kHz 8-bit µ-law samples in the flash sample bank. This is deliberately more compact than the embedded 16-bit PCM header and lets the firmware jump to slice points or play backwards without loading whole files into RAM. The standard 2 MB build reserves about 1 MB for uploaded samples; the 16 MB build reserves about 14 MB.

The web page converts source files to mono 24 kHz µ-law and shows how much of the sample bank they will use before upload. If the sounds are too large, the page will ask you to shorten them or use the 16 MB build. Save prepared sounds downloads the converted .pbank file, which can later be dropped back onto the page without reloading the original WAV or AIFF files. Factory fallback WAVs and VocalSamples.h are backed up in factory-samples/ for maintainers.

The folder picker is intentionally general: it looks for visible .wav files and ignores other files, including hidden macOS sidecar files such as ._Sample.wav. File names do not need to match the factory samples. Pick any four detected WAVs, audition them in the browser, and assign them to the four Punk Confusion slots. This can be useful for Squid Salmple, 1010music, Elektron, or Ableton export folders, but the page still converts the chosen WAVs into Punk Confusion's own .pbank format before sending them to the card.

Jack Map

LED Map

• LED 1: APC mode indicator in Switch Up; venue-zone display in room mode.
• LED 2: Broken Venue / vocal mode indicator.
• LED 3: divided APC trigger-clock blink in Switch Up; venue-zone display in room mode.
• LED 4: vocal trigger / Switch Down flash.
• LED 5: APC gate-open indicator in Switch Up; venue-zone display in room mode.
• LED 6: clip / chaos indicator.

Middle-mode venue display on LED 1 / LED 3 / LED 5:

While Switch Down is held, LED 1 / LED 3 / LED 5 temporarily become a three-step meter for the saved vocal-call trim.

Building

This release builds two beta firmware targets from the same source: one for standard 2 MB cards and one for 16 MB cards.

cmake -S . -B build
cmake --build build -j2

The generated UF2s are:

• build/punk_confusion_2mb.uf2
• build/punk_confusion_16mb.uf2

The firmware uses set_sys_clock_khz(192000, true), PICO_XOSC_STARTUP_DELAY_MULTIPLIER=64, and the default flash binary type. The 2 MB target reserves 1 MB at the top of flash for user samples; the 16 MB target reserves 14 MB.

To run the WebSerial loader page locally:

make webui

Then open http://127.0.0.1:8765/ in a Chromium-based browser.

Credits

• Card concept, samples, hardware testing, and release direction: Adrian Vos.
• Firmware and documentation assistance: Codex.
• ComputerCard library: Chris Johnson, MIT licensed.
• Workshop Computer platform: Music Thing Modular and Tom Whitwell.

This card intentionally builds against the local ComputerCard.h copy in this release folder. The upstream Demonstrations+HelloWorlds/PicoSDK/ComputerCard/ComputerCard.h copy is left untouched.

License

Punk Confusion code, documentation, and included vocal samples are released under the MIT License. See LICENSE.