MODULE / 0.1.26
Drop a .mid file anywhere here. Everything runs locally.
Master output / EQ & compressor
Shapes the final stereo output. Safety high-pass and soft ceiling remain active when processing is bypassed.
Effects / reverb, echo & chorus
FX mix controls all effect returns. MIDI CC91 sends to reverb; Echo adds repeats independently. CC93 sends to chorus; the Chorus knob adds a global send. Stereo width affects wet returns only. Synced echo follows tempo changes and playback speed (maximum 1.5 s). Room changes crossfade. Clear tails keeps held notes playing.
Renders directly (not through playback), then encodes locally. MP3 and OGG / Vorbis work offline in this file.
Plays a short phrase and pauses song playback. Your loaded MIDI and position are kept.
Changelog — v0.2.9
WASM DSP roadmap
Native Web Audio oscillators, filters and automation remain the efficient path for ordinary subtractive patches. WASM development targets shared physical state, feedback systems and nonlinear per-sample processing that standard nodes cannot represent directly.
- Oversampled nonlinear and FM renderer — in progress: v0.2.8 added 4× oversampled two-operator FM attacks with per-sample index envelopes. v0.2.9 adds a 4× oversampled soft-saturation path for generated electric-guitar buffers. Pickup coloration and polyBLEP or integrated saw generation remain planned. Native oscillators remain the default for simple pads and leads.
- Extended string core: one or two loop-dispersion allpasses plus light velocity-driven tension modulation. Preserve the existing cold-generation budget and three-mode fallback.
- Shared modal bank: one age-resumable mode summer for piano/soundboard coupling, mallets, toms, timpani and cymbal residuals, with ratios, RT60 values and excitation vectors supplied by recipes.
- Feedback wind family: stable reed/lip waveguides and a flute jet-delay variant driven by velocity and CC2. Current source-plus-filter winds remain the fallback whenever a feedback model becomes unstable.
- Live modulated FDN room: promote the existing eight-line FDN to a continuous wet processor with slow delay modulation and retained tails when AudioWorklet support is safe. Keep generated convolution impulses for file:// compatibility and offline verification.
Aftertouch and sostenuto support follow the modal and wind engines that can consume them. MIDI parsing, mixer UI and basic subtractive recipes intentionally remain JavaScript/native Web Audio work.
About this adequately ambitious prototype
48 melodic recipes span additive, subtractive, FM, modal, plucked-string, ensemble, wind-inspired and formant synthesis. GM programs without their own patch use a clearly labeled family approximation. Channel 10 uses 47 procedural drum recipes. No recordings, SoundFonts, fonts, downloads or dependencies.
Small channel meters show post-mixer audio level before effects, with a short decay. The trim option skips only empty end-of-track tails longer than 30 seconds; event timing and the original declared file length are preserved.
Choose a channel instrument to override the file, or select Follow MIDI. Approximate patches carry ≈ labels. Overrides survive pause, seeking and MIDI resets; another song or Reset mixer clears them. Changing an override during playback rebuilds held notes at their current age with a brief restart. Channel 10 stays percussion.
Play unlocks browser audio. Seeking and resuming rebuild controllers and restore the age of held notes. Oscillator phase and past effect tails restart. Background playback can be affected by browser timer throttling. GM2, GS/XG, pressure, bank-select sound sets and MIDI hardware input are future work.
Volume and pan sliders are mixer offsets applied alongside MIDI controllers. MIDI CC91 adjusts reverb sends and CC93 adjusts chorus sends. FX mix controls all wet returns. Echo and global chorus have separate controls in the Effects panel. Mute and solo gate new effect sends; existing tails decay naturally. Clear tails discards those tails. Stop clears the position; Pause keeps it.
Diagnostics include parser fixtures, GM state checks and procedural audio rendering.
Project license
Original MIDIogre code is
released under the
Unlicense. Any incorporated third-party portions retain their own licenses,
notices and attribution requirements.
Incorporated or adapted source code
None in
this build. The parser, synthesis engines, recipes, effects and
interface are original implementations. The projects below were
researched as behavioral, algorithmic or design references; their
source code, presets and assets are not embedded.
During unusually dense passages, some uncached string notes use a lighter procedural tone to keep playback responsive. Cached notes retain the full string engine.
String-model concepts also reference Julius O. Smith's Physical Audio Signal Processing and Jaffe/Smith's extended Karplus–Strong work. These are copyrighted research references; no book text or implementation code is incorporated.
Research and design references
| Project | Upstream license | Reference area |
|---|---|---|
| JZZ | MIT | MIDI messaging, SMF ecosystem |
| Tone.js | MIT | Audio-clock scheduling, envelopes, synth composition |
| SpessaSynth / spessasynth_lib | Apache-2.0 | sequencer, channel/controller separation |
| WebAudioFont | GPL-3.0 | player/envelope API |
| html-midi-player | BSD-2-Clause | piano-roll interaction |
| WebAudio Controls | Apache-2.0 | touch-friendly sound-module controls |
| Butterchurn | MIT | analyser-driven visualizers |
| Faust | GPL-2.0 compiler; architecture/library components have their own terms | DSP graph/compiler |
| Faust libraries | Varies by library; consult individual file headers | physical models, filters, effects |
| FluidSynth | LGPL-2.1 | GM channel behavior, voice lifecycle |
| TiMidity++ | GPL-2.0-or-later | SMF playback, GM behavior |
| Dexed | GPL-3.0; msfa engine Apache-2.0 | FM operators/envelopes |
| Surge XT | GPL-3.0 | oscillator/filter/voice architecture |
| Helm | GPL-3.0 | polyphony/modulation/UI |
| ZynAddSubFX | GPL-2.0-or-later (verify components before reuse) | additive/subtractive/pad engine separation |
| WebAudio TinySynth | Apache-2.0 | compact procedural GM timbre recipes, node-based synthesis |
| ks-guitar-synth | MIT | Karplus-Strong browser synthesis |
| STK | STK custom permissive license | plucked/wind/modal model taxonomy |
Licenses can vary by component. Upstream project links provide the authoritative license files. These are optional reference links; opening them requires Internet access, while MIDIogre itself remains offline-capable. Architecture, provenance, diagnostics and limitations are documented in the source.
Read MIDIogre's Unlicense
This is free and unencumbered software released into the public domain. Anyone is free to copy, modify, publish, use, compile, sell, or distribute this software, either in source code form or as a compiled binary, for any purpose, commercial or non-commercial, and by any means. In jurisdictions that recognize copyright laws, the author or authors of this software dedicate any and all copyright interest in the software to the public domain. We make this dedication for the benefit of the public at large and to the detriment of our heirs and successors. We intend this dedication to be an overt act of relinquishment in perpetuity of all present and future rights to this software under copyright law. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. For more information, please refer to <https://unlicense.org/>