FILTRE

A pocket French-touch sampler for the M5Stack Cardputer-ADV + TERM512 lid. Chop a record onto the keys, sweep one filter, let the 909 kick pump it, arrange the song and bounce mix + stems to SD. Guide: docs/FILTRE.md.

Study a reference track bar by bar and load its arrangement into the song page: the arranger.

FILTRE is a fork of Microgroove by lebiro.studio (MIT), which itself derives parts from Cardputer-Adv-Tracker by qwertyuu (MIT). Everything below this line is the upstream Microgroove README, kept for the instrument it still contains.

---

Microgroove

A wallet-sized groovebox that turns $30 of hardware and a 3D printer into a four-track acid powerhouse.

Firmware for the M5Stack Cardputer-ADV, by lebiro.studio.

Print it. Flash it. Jam.

What it does

• 3 synth tracks — each switchable between a mono 303-style voice (saw/square/tri/sine/wavetable → resonant SVF filter, accent and slide) and 2–3-voice polyphony. Overlapping notes record a slide in mono, a chord in poly.
• 8 drum lanes — each independently 808 synthesis, 909 synthesis, or SD sample playback; per-lane volume, tune (±12 st), decay, and choke groups (909 open/closed hats choke by default).
• Live mic sampling — hold one key; the footer becomes a level meter. Release to auto-trim, write to SD, and play it on a drum lane instantly (max ~2.6 s).
• Resampling — hold SONG while playing to bounce ~1.9 s of the master mix onto any pad.
• Sequencer — 8 patterns × 16 steps, hold-to-clone, bar-quantized pattern switching, live record with quantize / step-write / hold-to-erase.
• Song mode — 64-slot pattern chain with loop point.
• User wavetables — drop single-cycle WAVs (AKWF) on the card; they appear as oscillators.
• 8 project slots on microSD; sampled sounds reload with projects by filename.
• One key = one function — hold any orange-labeled key 0.5 s for its second function, with a progress bar so nothing fires by accident. The note keys mirror a real piano layout, E–F and B–C gaps included.

Full feature tour and guides: docs/USER_MANUAL.md

Get it running

Pick whichever flashing route suits you — all three are on the Releases page.

Option 1 — pre-built binary: grab microgroove.bin (a merged image) and flash it to offset 0x0 with esptool (esptool.py write_flash 0x0 microgroove.bin) or any ESP32-S3 flashing tool.

Option 2 — Arduino IDE (from the release zip): download Microgroove_source.zip, unzip it (keep the folder named Microgroove), open Microgroove.ino, install the M5Cardputer library (pulls in M5Unified/M5GFX), select the M5Cardputer board (or ESP32S3 Dev Module with USB CDC on boot), and Upload. A HOW_TO_FLASH.txt is included in the zip.

Option 3 — build from source (this repo): clone it and open Microgroove.ino in the Arduino IDE, or use PlatformIO:

[env:m5stack-cardputer]
platform = [email protected]
board = esp32-s3-devkitc-1
framework = arduino
build_flags = -DESP32S3 -DARDUINO_USB_CDC_ON_BOOT=1 -DARDUINO_USB_MODE=1
lib_deps = M5Cardputer=https://github.com/m5stack/M5Cardputer

Then, for the demo + samples: copy either the release's Microgroove_SD_card.zip contents or factory-sd/groovebox/ to the root of a FAT32 microSD. Power on → hold LOAD → tap SONG → PLAY. No card? Hold LOAD+SAVE together for a built-in demo.

The shell

Print files, keycap label sheet (v6), and assembly guide: MakerWorld · hardware/

Keys at a glance

T1 T2 T3 TD select tracks (hold = mute) · P1–P8 select patterns (hold = clone) · the note keys are a piano (whites on the home row, sharps above, E–F/B–C gaps dead), and the first eight white keys become pads 1–8 when the drum track is selected. Orange = the hold (0.5 s) function; green = the sampling holds. AUX held samples the mic, SONG held while playing resamples the mix. That picture is upstream Microgroove; FILTRE moves several keys. Hold ok for the current map on the lid, or read the FILTRE guide.

Architecture

Microgroove.ino     setup / main loop (core 1: input, sequencer, UI)
audio_engine.cpp    render task (core 0), dual buffer -> Speaker.playRaw
synth_voice.h       303-style voice (osc + SVF + envelopes)
sequencer.h/.cpp    SynthTrack (1-3 voice alloc), patterns, transport, live record
drum_voice.h        808/909 synthesis + per-lane engine/choke logic
sampler.cpp         WAV decode -> 192 KB RAM pool, playback voices
mic_sampler.cpp     mic capture + engine resampling -> SD/pool
wavetable.cpp       8 built-in tables + user single-cycle WAVs
storage.cpp         GBX v2 project files (loads v1 transparently)
input.cpp           keyboard snapshot diffing -> short/long-press dispatch
ui.cpp              5 pages, sprite double-buffered
keymap.h            every key assignment in one file

All firmware files live in the repo root next to Microgroove.ino (Arduino sketch layout) — open the folder in the Arduino IDE and it picks them all up.

22.05 kHz, 256-sample buffers, all voices rendered per-sample into a soft-clipped mix on core 0. Project files are versioned: v2 stores chords and per-track VOICES; v1 files load transparently (and become v2 on the next save).

License & lineage

MIT — see LICENSE. The synth voice, 808 drum synthesis, and audio task architecture are derived from Cardputer-Adv-Tracker by qwertyuu (MIT), substantially redesigned into a different instrument. The factory sample pack is CC0 by lebiro.studio.

If Microgroove earns a place on your desk: Ko-fi ☕