VIBE FM Mk1 Open firmware, officially supported. The first 100 get 10% off Join the list →

The Firmware Hub · FM-1 · FMav-1

FMav-1 for the M-VAVE FM-1

FMav-10.6.4 · beta

by wootrop & gpix GitHub ↗

synthbeta

A semi-modular mono synth: VCO, self-oscillating 4-pole ladder filter, LFO, ADSR, VCA, wave folder, S+H, mixer and a patchbay with normals and dummy plugs. Four tracks on their own MIDI channels, MIDI clock sync, arpeggiator, chorus, delay and reverb, 90 sounds. Made for blind players too.

Text by the maker.

About it

Runs on
FM-1
Release
0.6.4 · released 20 Nov 2026 · on the hub since 16 Nov 2026598 KB · what's new
Licence
GPL-3.0 · source ↗ · copy on the hub
Flash chip
Every FM-1, the newer chip 0xB36014 too
Identity
FM-1_3628what it reports over USB while it runs
Build
Built by its maker, not signed
Popular
↓ 9 installs via the hub

Family

Like FMav-1

Changelog

What changed, version by version14 versions

0.6.4 (2026-11-20)

  • Screen reader: a section's button held reads its page's knobs by name only, no values ("1 RATE, 2 WAVE, 3 SYNCHRONIZATION, 4 RATE FINE"; HOME's mixer: PLAYING after a track that sounds). A value: REC held and its knob turned.
  • Screen reader: a knob with nothing on it (an empty place on the page) says "NOT ASSIGNED" at every turn.
  • Screen reader: the FM-1's own voice on at power-on says the firmware and its version during the start-up picture, "F MAV ONE VERSION 0 POINT 6 POINT 4", then the program and the page.
  • Screen reader: SELECT turned says the page at once, at every detent (it waited for the turn to stop); at the first or the last page, turned on, again.
  • The FM-1's own voice at power-on: the program and the page wait for "F MAV ONE VERSION ..." to end (they cut its last words, and a knob turned cut it all). A warning (kept whole) is no longer cut in its last sounds either.
  • Screen reader, the patchbay: the inputs' pages are "INPUTS" alone ("2 OF 4. INPUTS", not "INPUTS 1 TO 4"); a cable is "LINKED TO" its input ("LFO LINKED TO 1 VOLT PER OCTAVE", "NO CABLE LINKED TO FOLD IN"), on its knob and in the patchbay's overview.
  • Screen reader: REC held says the track and the page first: "CURRENT TRACK 2, CURRENT PAGE 1 OF 2. OSCILLATOR. TURN A KNOB TO KNOW ITS VALUE".
  • Screen reader: a patch cable's knob says the whole phrase at every turn, not only the first: "LFO TO FOLD IN", "VCO TO FOLD IN", "NO CABLE TO S H NORMALIZED TO V C O" (the other knobs with few values: their value alone).
  • The FM-1's own voice turned on (PLAY held 2 s) says how to use it instead of the page: "SCREEN READER ON", a longer pause, "PRESS AND HOLD ANY PAGE NAVIGATION BUTTON TO KNOW THE PARAMETERS CORRESPONDING TO KNOBS 1, 2, 3, AND 4. PRESS AND HOLD THE LAST BUTTON ON THE SECOND ROW TO SAY THE CURRENT PAGE, AND LISTEN TO THE ON-SCREEN VALUE OF A PARAMETER WHILE ADJUSTING THE CORRESPONDING KNOB. HAVE FUN WITH F MAV ONE!" (a button or a knob cuts it short).
  • Screen reader: REC held, a continuous knob said its name twice (on the touch, then with its value): once now.
  • Screen reader: after a second with no knob turned, the next turn says the knob's name again, the same knob too (a knob with few values: its name and value).
  • Screen reader: REC held, a knob with few values says its name and its value at every turn ("DELAY ON"), as a continuous knob does with REC held.
  • Screen reader: the patchbay's DUMMY is said "DUMMY CABLE", always ("DUMMY CABLE TO FOLD IN"); the screen keeps DUMMY.
  • The FM-1's own voice: SPEED 6 at first (it was 4), on a first install or after a memory reset; saved settings keep their SPEED.
  • The Firmware Hub's description, shorter: "... 90 sounds. Made for blind players too."

0.6.3 (2026-11-19)

  • Screen reader: OCT- / OCT+ say the octave again, as their light shows it, the number first: "PLUS 2 OCTAVE" (the light: +-1 lit, +-2 blinking slow, +-3 fast), "MINUS 1 OCTAVE", and pressed on at the end, again; both together "0 OCTAVE".
  • The knobs glide: a continuous knob turned (or its MIDI CC) no longer moves the sound in steps. Its value glides to the new one in ~12 ms: CUTOFF, RESONANCE, PULSE WIDTH, the waves (they glided in ~3 ms), PITCH, FINE, the fines, LFO RATE, ANALOG, KB SCALE, the mod mixes and amounts, VOLUME, ATTENUATOR, ONE LVL, FOLD, NOISE, SUSTAIN, the effects' sends, rates, depths, feedback, cuts, widths, decay, damping, diffusion, pre-delay, the mixer's PAN and GLOBAL TUNE. Not the times (ATTACK, DECAY, RELEASE, GLIDE, the gates: a speed, no step to hear), nor the switches, modes and cables. A program loaded comes at once (CUTOFF and the others too: they glided from the last program's); at rest every knob gives exactly its value (CUTOFF and the others stopped a hair under a value reached from below). The 90 programs sound as in 0.6.2 (a patch with S+H, noise or a singing filter never repeats sample for sample). The worst case's CPU: +1.3 % (+27 instructions a sample).
  • The knobs on the screen glide too: a knob turned slides to its new place (a quarter of the way every frame, every ~12 ms or more) instead of jumping; a page drawn anew shows them where they are at once.

0.6.2 (2026-11-19)

  • HOME > READER > OUTPUT (knob 4, the user): where the FM-1's own voice goes. MAIN (the default): with the music, on the outputs and in USB audio (0.6 kept it out of USB audio); HEADPHONES: the outputs only (the jack at the back); USB: USB audio only. The music is 6 dB lower while it speaks only where the voice is. No SPEAKER choice: the speaker and the jack get the same signal (one DAC), and no line is known that turns the speaker off, so the voice can't be on the speaker alone; with nothing in the jack, HEADPHONES is heard on the speaker. Kept in the settings (the byte that was spare: older settings start on MAIN).
  • A knob's value too wide for its cell even with tight letters (HEADPHONES) is drawn in the small letters.
  • Screen reader: an input with a normal (ONE, S+H, ATTN) says what it is normalized to only when its knob is on NO CABLE ("NO CABLE TO S H NORMALIZED TO V C O"); with a cable in, only its name ("LFO TO S H"), in the patchbay's list of cables too.
  • REC tapped: the screen's EFFECTS ENGINE ON / OFF shows the space between EFFECTS and ENGINE (the message's letters have a 2-pixel space); said only "EFFECTS. ON" / "EFFECTS. OFF".
  • The Firmware Hub's description says FMav-1 is made for blind players too.
  • Screen reader, a discrete knob (25 values or fewer: a switch, a mode, a cable, BEND, OUTPUT): each value, the value alone (its name only with the first value once it is touched: "DELAY OFF", then "ON"; a cable "LFO TO FOLD IN", then "NOISE", "NO CABLE"), never MINIMUM / MAXIMUM; turned on against an end, its value again. A continuous knob (more values) says its name when touched and, the moment it reaches an end, "MAXIMUM, 19 KILOHERTZ" ("MINIMUM, ..."), without its name.

0.6.1 (2026-11-19)

  • Screen reader: a mix at its middle ("50/50": VCO WAVE, LFO WAVE, VCO and VCF MOD MIX) is said "50 50", still heard "fifty fifty" on both voices: fm1-reader's audit took "fifty fifty" for a word said twice. The audit now finds no errors; its allow list (tests/sr_allow.txt) has the words FMav-1 means that its dictionary lacks (ATTENUATOR, OMNI, CENTRE, PRE-DELAY, PATCHBAY, INIT and three programs' names). The words file is the same as 0.6's.

0.6 (2026-11-19)

  • Screen reader (issue #1), two ways, the same words: on a computer, the page https://charlesvestal.github.io/fm1-reader/ (Chrome or Edge, the FM-1 on USB: its own voice or VoiceOver, NVDA, JAWS, TalkBack); or the FM-1's own voice, no computer: PLAY held 2 s turns it on ("SCREEN READER ON", then the page) or off ("SCREEN READER OFF"), kept after a power-off. What is said:
  • on connecting or turning the voice on: the program and the track, then the page;
  • a new page (a section's button, SELECT): its place among its button's pages and its title only ("1 OF 7. DELAY"; the VCO's "OSCILLATOR", the VCF's "FILTER", the arp's "ARPEGGIATOR", without a state or a mode; a button with one page: the title); the patchbay's overview, its cables ("LFO TO PWM"); the same button again on a one-page section says it again;
  • a knob touched, the first on the page or another than the last: its name at once ("L F O RATE"; a switch: its ON / OFF);
  • a section's button held half a second (as long as REC for its CCs): its page with every knob's value (the buttons act on the release now, as EDIT and GLO did); HOME's mixer: PLAYING after a track whose meter shows;
  • a knob of few values (25 at most: a switch, a mode, a cable "LFO TO FOLD IN", a note value, BEND): each value as it changes; any other knob says nothing while it turns, but its value when turned with REC held (once set) and its value and MINIMUM / MAXIMUM the moment an end is reached ("CUTOFF 19 KILOHERTZ, MAXIMUM"); PRESETS the program (or the slot SAVE writes to), ALGORITHM the track, SELECT the page, once they stop; not the octave (playing tells); a knob turned stops the page's reading (said or still to come);
  • every message on the screen (KEYBOARD HOLD, the effects, SAVE, undo, the reset's question); a warning heard whole;
  • never a MIDI CC, never the hint at the bottom right of the screen. The short words in full (the user's list): TIME L "time left", KB "keyboard", EG "envelope", AMT "amount", 1V/OCT "1 volt per octave", PRIO "legato priority", LIGHTS "lights brightness", GLIDE "portamento", SYNC "synchronization", BEND "pitch bend", FOLD "wave fold", ANALOG "analog effect", TUNE "global tune", ONE (-5) "one normalized to minus five", the chorus TYPE 1, 2, 1 plus 2, free; a switch by its whole name ("VCA MODE SWITCH"). One place makes every text (mfm_speak.h) and both ways say it, so they never differ: the tests check every word said is in the voice's dictionary and the page says them word for word. fm1-reader's shared code (announce.c, tools/gen_words.py, MPL-2.0) is fm1-base's, by Charles Vestal; silent until a reader asks (a DAW on the same port never sees a byte). The voice is Bigorna's formant synthesiser (Klatt, the DECtalk kind; Nuno Zimas, GPL-3.0), made on the FM-1 at 22 kHz, heard on the outputs (the music 6 dB lower while it speaks), not in USB audio; it pauses on commas and stops. About 20 KB of flash and 2 KB of RAM; while it speaks, ~140 instructions a sample on the host (about 5 points of the worst case's CPU).
  • HOME > READER (the user, the last of HOME's pages): the voice's VOLUME (0-10, on the MASTER knob's level), SPEED (1-9: 1 the slowest, 4 at first), PITCH (80-200 Hz); the page shows, big, "FOR ACTIVATE THE SCREEN READER, HOLD PLAY/STOP".
  • REC held (the user): "TURN A KNOB TO KNOW ITS VALUE" said; each knob's MIDI CC in a badge over its middle, its value left in view (the patchbay: over each input's jack); a knob turned with REC held doesn't turn the effects off when REC is let go.
  • PLAY: a tap is HOLD as before, now on the release; the message is KEYBOARD HOLD ON / OFF.
  • REC tapped: EFFECTS ENGINE OFF / ON (said "EFFECTS. ENGINE OFF").
  • SAVE says "SAVE PRESET, ROTATE THE PRESET KNOB FOR SELECT THE SLOT, SAVE AGAIN FOR CONFIRM"; held 0.6 s (no key yet), the keys it opens ("F3 INIT, G3 RANDOM. A3 MUTATE, G5 RESET").
  • ARPEGGIATOR (the user): knob 1 is its ON / OFF switch (a new parameter, CC 89 / NRPN 72), MODE has only the modes (UP .. RANDOM; CC 55 too), GATE on a second page (ARP again). A program saved before keeps its arp: MODE OFF becomes the switch off, a mode the switch on with that mode.
  • TEMPO turned under an outside clock says what didn't happen: "TEMPO NOT CHANGED, THE USB CLOCK SETS IT".

0.5.2 (2026-11-18)

  • ATTACK as the reference's (the user, by ear: the reference's ATTACK at 10 o'clock sounds as 0.5.1's noon): 153 ms at 10 o'clock (it was at noon); 44 ms at 9 and 1.15 s at 3 (the scope's) stay; noon is ~350 ms now. The factory programs and the programs saved by 0.5 / 0.5.1 keep their attack time (ATT_FROM_051; the saves' tag is MFM7 now): only the knob's position moves.
  • DECAY 1.2 s at noon (the user; was 668 ms); 437 ms at 9 and the ends stay. The programs keep their decay time as above (DEC_FROM_051).
  • RELEASE 458 ms at 9 o'clock (the user; was 330 ms); noon 627 ms and the ends stay; the programs keep their release time the same way (REL_FROM_051). No program changes its sound.
  • REC held shows the MIDI CCs (the user): after half a second every knob on the screen shows its CC in a filled badge (CC 74) instead of its value, the PATCHBAY each input's CC instead of its name; NO CC for the globals. Let go: the values again, the effects untouched. A short tap (under half a second) still turns every effect off / on, now on the release.
  • CC 7 is the track's mixer LEVEL (the user: the four levels with a CC): on each track's channel it moves its LEVEL knob, on its scale (0-127: 100 = unity, 80 the default with four tracks), shown and saved with the settings. It was a separate gain multiplying the LEVEL, unseen.
  • CC 10 is the track's mixer PAN (the user): 0 all left, 64 the centre, 127 all right; shown and saved the same way.

0.5.1 (2026-11-16)

  • KEYBOARD moved to SELECT pressed (the user), its second page after UTL; HOME is the MIXER (levels, pans), VOICE and GLOBAL.
  • A new patch source, GATE: +5 V while a note is held (the FM-1's keys or MIDI, the arpeggiator's notes too), 0 V when let go. It takes the slot the CLOCK output left in 0.5, so no other cable's CC value moves (GATE: 109); a cable from CLOCK in a program saved by 0.4.1 or before still becomes no cable.
  • PRESS is called AFTERTC (the MIDI aftertouch).
  • The external sources' cables (VELO, MODWHL, AFTERTC, GATE) share one colour, khaki, apart from the panel's outputs; two coming into one column enter from their own points and bow apart.

0.5 (2026-11-16)

Every knob calibrated against the reference instrument (the user's recordings, scope captures and photos of its panel), the 90 programs rebuilt from INIT so they can be set up on it, and the panel reorganised.

  • The knobs turn as the reference panel's: drawn from 7 to 5 o'clock (300 degrees, was 270), eleven ticks on the hours (the big knobs: a tick every 10 degrees). A knob set at 2 o'clock on the panel is set here at its 2 o'clock tick. Every law is in one table, KNOB at the top of tools/gen_tables.py (a value at 7 o'clock, at any hours in between, at 5 o'clock, and "exp" or "lin").
  • A more faithful ladder: each stage saturates its own state as well as its input, as the transistor pairs do (dV/dt ~ tanh(Vin) - tanh(V): Huovilainen 2004, D'Angelo & Valimaki 2013); ~8 % more CPU in the worst case. Fitted to the reference's recordings (its INIT, a C2 saw, five CUTOFF / RESONANCE settings): the -3 dB corner at ~24, 289 and ~4800 Hz at 9 o'clock, noon and 3; the whistle at 2.12 times the corner; RESONANCE starts to whistle halfway between 3 and 4 o'clock. Checked on the FM-1 itself (its USB audio): the harmonics within 0.7 dB open and 1.3 dB at RESONANCE noon, the whistle 589 Hz (the reference's 583). CUTOFF at 10:28 is 95 Hz (a second recording).
  • The ladder sings from silence, as the hardware's: past the singing point it starts with nothing in it (a hiss of one 16-bit step a block); under it, silence.
  • CUTOFF fully counterclockwise is silent, with no click (the pulse's DC no longer reaches the VCF).
  • The EG: ATTACK 0.8 ms .. 2.5 s on an RC curve aiming at 1.11 times the top, restarting from 0 at each new gate as the reference's; DECAY 3 ms .. 16 s and RELEASE 3 ms .. 18 s in the reference's seconds; at noon ATTACK 153 ms, DECAY 668 ms, RELEASE 627 ms; at 9 o'clock ATTACK 44 ms, DECAY 437 ms, RELEASE 330 ms; SUSTAIN 13 % at 9 o'clock, 48 % at noon, linear above.
  • VCF MOD AMT follows |x|^1.5 (was the square: 3.3 octaves at 2:48).
  • The LFO: 0.165 .. 550 Hz, 4.7 Hz at noon; between triangle and square as the reference's (the square high while the triangle rises: halfway it steps at the peaks), both skewed 58 / 42.
  • The pulse width runs 0 .. 50 % (knob, PWM AMT and the PWM jack; it was 1 .. 99 %): fully right is always a square, and a modulation under 0 % silences the pulse, as the reference does.
  • S+H IN's normal is the VCO again, as the reference panel says (S+H (VCO)); programs saved by 0.4 .. 0.5's first builds that sampled the noise get a NOISE > S+H IN cable and sound as before.
  • The CLOCK output is gone (the reference panel has none); LFO SYNC, the arpeggiator and the delay still follow the clock. A saved cable from CLOCK becomes no cable; CC 84 and 85 do nothing.
  • The 90 factory programs rebuilt from INIT: each lists only the knobs its sound needs (every other one at INIT's where the simulator heard no difference) and uses only what the reference panel has, the effects aside (no white noise, no VELO / MODWHL / PRESS cable, no FINE, ANALOG, arp or LFO SYNC), so each can be set up on the reference from its INIT by turning only the knobs listed (firmware/src/mfm_patches.h, in the panel's order). Their sounds are 0.4.1's, each knob moved to the same Hz, time and level on the new curves, the levels as 0.4.1's. The noise programs (37 NOISE HIT, 69 CLOCK HATS, 79 WIND, 81 SNARE, 82 HI-HAT, 83 CLAP) make their noise with the VCO and the LFO modulating each other at audio rate; the rhythmic ones have their LFO free at the rate it had at 120 bpm; 42 ARP PLUCK and 70 SYNC ARP no longer arpeggiate, 59 WHEEL LEAD no longer takes the wheel.
  • A new order: 01 BRASS LEAD, 02 CLASSIC LEAD, 03 CLASSIC BASS, 04 PWM LEAD, 05 FOLDED TRI, 06 DOTTED LEAD, then the others in their old order. The four tracks start on 01, 11, 21, 31 (first power-on, the memory's reset).
  • CLASSIC BASS on the reference's own knobs (a photo, and the laws fitted to its recording); BRASS LEAD with the user's EG and LFO RATE.
  • INIT (91-99, SAVE + F3) as the reference's: the saw, PULSE WIDTH 50 %, both MOD MIXes on EG, the LFO a triangle, CUTOFF open, RESONANCE 0, ATTACK, DECAY, RELEASE 0, SUSTAIN full.
  • The buttons reorganised: HOME opens the MIXER (levels, then pans), then KEYBOARD, VOICE, GLOBAL; EDIT the VCO; GLO the VCF; the PERFORM pages are gone.
  • Deterministic navigation (for the blind, and muscle memory): a section's button opens its first page, pressed again its next one (pressed N times = page N); SELECT stops at the first and the last page; EDIT and SAVE act on their release however long they were held. The guide has the map of every page.
  • FX: the delay first, then the reverb, then the chorus.
  • CUT FINE (GLO, the VCF's third knob) and RATE FINE (LFO, fourth knob): +-2 semitones of fine tuning (CC 87 / NRPN 70, CC 88 / NRPN 71). Not on the reference panel.
  • ARP pressed again no longer turns the arp on / off: its MODE knob does.
  • The memory's reset: SAVE held + G5 asks "RESET ALL MEMORY?", OCT+ confirms (anything else, or 8 s, cancels): every saved program and setting goes and the FM-1 restarts as on its first install.
  • The LEDs: HOLD lights PLAY green; the LFO button no longer blinks with the LFO (its LED is on the screen); fixed: REC's LED (the effects on) stayed dark after power-on.
  • The patchbay cables bow across their line: a cable between two jacks of one column no longer runs straight over the others (the sides alternate), and parallel cables never coincide.
  • The version on the boot screen.

0.4.1 (2026-11-14)

  • The clock's source: GLOBAL > CLOCK = INTERNAL (the TEMPO knob; every MIDI clock and Start / Stop is ignored), USB or TRS (the clock and Start / Stop / Continue only from that one, the other ignored). Until 0.4 both counted: a sequencer sending its clock on USB and TRS at once doubled it. A fresh start (or an update from 0.4) is INTERNAL.
  • With USB or TRS the arpeggiator follows that clock only: its steps on the clock's ticks; with the clock stopped (or none yet) it plays the first note and waits, never at the TEMPO knob's pace.
  • With USB or TRS TEMPO is locked (grey; turned, it says so) and shows the tempo the sync reads (TRS 125; --- until measured). The tempo is measured after one quarter already (then refined every four, as before).
  • GLOBAL's knobs: MIDI CH, TUNE, CLOCK, TEMPO, then KEY VEL, LIGHTS. The CLOCK page says USB CLOCK / TRS CLOCK, (STOPPED), or NO CLOCK.
  • The sound is unchanged.

0.4 (2026-11-14)

  • White noise: a generator a track; a NOISE level mixes it into the VCF's input (with the VCO or the folder), and a NOISE output on the patchbay takes it to any input. With DUMMY in FOLD IN the VCF hears the noise alone.
  • The S+H samples the white noise when S+H IN has no cable (it sampled the VCO). Programs saved by earlier versions that use the S+H get the cable VCO > S+H IN, so they sound as they did.
  • No knob twice but on PERFORM: the cables only on SEQ (the modules' INPUTS pages and the S+H page are gone), UTL one page (FOLD, ONE LVL, ATTENUATOR, NOISE), TEMPO only on GLOBAL.
  • A section's button pressed again turns to its next page, as SELECT (ARP again: the arp on / off).
  • 18 factory programs remade with the noise (drums, wind, the random S+H sounds, a breath on a few leads and pads, a noise riser); the other 72 render bit-identical to 0.3.1.
  • The programs' name has a fixed place (new parameters need no conversion from now on); 0.1 .. 0.3.1 saves converted.
  • The patch cables' CC values change (one more source: see CC-NRPN.md); NOISE is CC 86.

0.3.1 (2026-11-14)

  • DUMMY last in every input's list (after the MIDI sources).
  • MIDI channel 1 by default (tracks 1-4 on channels 1-4) on a fresh start; a saved channel stays.
  • SEQ has two items: PATCHBAY (every cable) and INPUTS (the 12 inputs on knobs, 3 pages); SEQ pressed again switches between them.
  • The keys a held combination can press are lit: SAVE held lights F3 / G3 / A3 (INIT, RANDOM, MUTATE), EDIT held lights OCT- / OCT+ (undo / redo).
  • The sound is unchanged: the 90 programs render bit-identical to 0.3.

0.3 (2026-11-13): FMav-1

  • The name: FMav-1, identity FM-1_3628 (0.1-0.2 had a working name and FM-1_988, which is another firmware's on The Firmware Hub). Programs saved by 0.1 and 0.2 are converted.
  • REC: every effect off / on together, each sound's settings kept (lit = the effects play; remembered).
  • The MIDI clock (TRS or USB) as a source: LFO SYNC (FREE, 8/1 .. 1/32T, phase-locked, START realigns), a CLOCK output (note value, GATE length) any input can take, the arpeggiator on the clock's grid. TEMPO without a clock.
  • DUMMY cable on every input (the normal broken, 0 V), one detent after "---".
  • The mult is gone; VCA OUT is the audio out (no cable from it). Old cables from the mult are taken from its source.
  • SEQ: the whole patchbay first (full screen, external MIDI cables from outside the panel, a dummy hanging loose), then all the inputs on the knobs with a list of their cables. SELECT pressed opens UTL (folder, mixer, attenuator); no page has a switch on SELECT any more (VCA MODE and the effects' ON are knobs).
  • 90 factory programs (50 new); programs 1-40 sound bit-identical to 0.2.
  • OCT- / OCT+ LEDs show the octave (lit, slow blink, fast blink).
  • The LFO's lights are steady when the LFO is faster than they can show.
  • Fixed: a lit knob erased part of its box's border; MIDI-source cables were drawn with a plug at their far end.

0.2 (2026-11-13)

Four tracks; chorus / delay / plate reverb as shared sends; an animated UI of our own drawing; fast knob turns counted (the FMini-1 #126 decoder); INIT = the reference manual's initial patch; jacks never offer their own module's outputs.

0.1 (2026-11-13)

First version: one track, the whole voice and patchbay, 40 programs.

CHANGELOG.md on GitHub ↗