Anharmonic Studio

A native desktop music studio for sampling, beatmaking, synthesis, arranging, vocal production, and mixing. Import a sound, cut it into playable slices, build a rhythm, write notes, arrange a song, and export a stereo WAV.

Explore the studio and watch the film

Get Anharmonic Studio

Official Linux, Windows, Intel Mac, and Apple Silicon Mac downloads fund development. Official released binaries are the supported distribution for musicians: download, install, and open the studio. They bundle the runtime and dependencies, with no compiler setup or subscription requirement to keep using your installed version.

The current packages are unsigned 0.1.0-rc.1 release candidates. Official packages intentionally remain unsigned and not Apple-notarized; paid signing is not a release requirement. Physical audio-interface acceptance remains unfinished. Read the unsigned installation guide. The standard download costs $1 USD once, plus applicable tax, through Stripe checkout. After payment confirmation, your selected installer downloads from private Cloudflare storage.

Option	Price (USD)	Includes
Source code	Free	Full source; build it yourself without an account or purchase
Official download	$1, one time	Current major version and its updates, on all supported platforms
Supporter download (planned)	$45 or more, one time	Current and next major version and their updates
Optional donation on Ko-fi	Any amount	Development support only; no download entitlement

The standard $1 download and separate Ko-fi donations are open. See downloads and pricing on the website and the distribution plan. Paid installers stay outside this public repository and Pages site. Source access stays free; recipients retain their GPL rights. The matching source and license packages for all four platforms are public GitHub release assets. Cloudflare R2 stores only the compiled installers.

Latest source updates

The October 4 source update begins the Qt Quick/QML interface migration with live workspace navigation: tabs on wide windows, a workspace picker on narrow ones, keyboard navigation, and a widget fallback if the scene fails. It uses the existing editors and production C++ DSP; project and audio state stay in their established owners. Transport overflow stays accessible in focus mode, audio setup preserves independent channel choices and rejects failed routes, and buffer durations are distinguished from measured round-trip latency. This is the first migrated interface surface, not a complete QML rewrite. Release self-checks now require the bundled QML scene to load.

The September 27 source update improves count-in recording, separate pad/key routing, Song recording, uninterrupted workspace switching, independent Native and Prism instruments per pattern, and undo/redo responsiveness. Prism now recovers from transient render backlog. See the changelog and recording guide.

These changes are newer than the paid 0.1.0-rc.1 packages. A new encrypted candidate build must pass validation before the private download catalog changes.

What is inside

Display-aware desktop: startup checks the local CPU/memory and the current screen's usable logical size before constructing the workspace. Laptop screens start with compact crates and transport controls, a wider editor and optional pads. Editors and application dialogs scroll when their controls need more room; display/work-area changes refit the window without changing your project.

Trap Foundry: 64 original, bundled 48 kHz / 24-bit drum one-shots, including 12 tuned 808s, kicks, snares, claps, closed/open hats, percussion and FX. They appear automatically in the Browser; choose Tools → Song / beat tools → Trap Foundry in the workspace to load an eight-voice kit and editable groove. New pads/Notes instruments use the 808s' known root notes. See the pack guide and license.

The current source includes a 128-track mixer, named timeline markers/regions, and audio-file analysis. See the usage guide and explicit limitations and full implementation backlog. These additions are not yet qualified for release or deployed to Cloudflare.

• Sampler: import audio, trim ranges, detect transients, and map slices to four 4×4 pad banks.
• Beats: program step patterns and control per-hit velocity.
• Instruments: explore factory sounds, shape analog synth patches, and record arpeggiator notes.
• Notes: write synth parts or play samples chromatically in the piano roll.
• Scoring: view a pattern or arranged song as engraved sheet music, select an individual instrument part, compose pattern notes, and export PDF or MusicXML. Notation uses 4/4, concert pitch and sixteenth-note rounding without changing playback. Audio → score automatically converts a file, selected library audio, or the current song into reviewable note parts. Song modes separate vocals, bass, drums and other instruments, with an optional six-stem mode adding piano and guitar. Single-instrument transcription runs with the bundled Basic Pitch model; song separation needs ./install-separation.sh and may download Demucs weights on first use. Review the estimated tempo and first four bars, select parts, then add editable patterns and muted score rows in one undoable operation. Conversion supports audio up to 15 minutes; fixed-tempo 4/4 notation and drum-family estimates may need correction.
• Song: arrange patterns and audio clips, create variations, and automate levels and pan.
• Mix: balance tracks with EQ, saturation, compression, delay, and reverb.
• Vocals: record directly into Song, open recorded clips in Autotune, build comps, and render pitch correction to a new take.
• Export: render the arrangement to a 48 kHz stereo WAV.
• Devices: automatically connect MIDI inputs, remember controller mappings, and load external VST3 instruments and effects.

The optional stem-separation engine requires extra dependencies and an initial model download. The core studio works locally after installation.

MIDI controllers and plugins

Open Tools → Devices & Plugins. MIDI inputs are checked every second and reconnected automatically. Choose Keys, Pads, or the combined mode (keys plus pads on MIDI channel 10). Use MIDI Learn to assign unfamiliar pad layouts, transport buttons, knobs, and faders. Mappings are remembered; unplugging a controller releases its held notes. MPK/MPC compatibility depends on the device exposing a standard MIDI input; proprietary modes may require the manufacturer's driver or a MIDI/controller-mode setting.

The plugin scanner checks standard installation folders and folders you add. Load a VST3 instrument for synth notes, and use Plugins → Manage insert plugin chains to configure serial effects on tracks, buses and the master. Parameters and presets save with the project and are used during WAV export. Pitch bend and unmapped MIDI CCs reach the external instrument during live playing. Arbitrary native plugin windows, MIDI output/clock synchronization, and recording expression automation are not implemented yet. Plugin controls use the studio's parameter panel; Prism has a dedicated Studio editor. Live parameter changes update a running instance where supported. Insert Native or Prism from the Instruments workspace to give each pattern its own instances; live input follows the selected instrument, and Layer current notes explicitly copies notes for stacking.

Plugins must match your operating system and processor. VST2, CLAP, and LV2 are listed but cannot be loaded by this host. Audio Unit support is available through the host on macOS; this integration has only been tested on Linux. Linux checks include the Nekobi instrument and MVerb effect; other plugins may need different bus layouts or host features. External instruments and effect chains run in isolated processes. Transient instrument backlogs recover while preserving queued MIDI releases. A crashed or timed-out instrument is silenced and a failed effect path is bypassed; failed exports report the error. A serial insert chain shares one two-buffer live bridge (about 21 ms at 48 kHz / 512 frames), in addition to plugin-reported latency and the native output queue.

The Audio interfaces tab refreshes available outputs, including Linux PipeWire devices. Choose the desired routing in Audio setup; connecting hardware does not automatically change your recording input or output.

Free source for developers

The full application source is free under GPL-3.0-or-later, including the scripts used to compile and install it. No purchase is required to study, build, modify, or share it under that license.

Source builds are intended for developers who maintain their own toolchain, dependencies, and audio configuration. We do not provide installation or configuration troubleshooting for self-compiled copies. A source checkout does not include a signed official installer or managed update channel. Reproducible application bug reports and patches are welcome.

• Developer build guide: prerequisites, native compilation, validation, and packaging for Linux, Windows, Apple Silicon Mac, and Intel Mac.
• Support scope: official builds and self-compiled copies. Purchase and installer help: [email protected].
• Contributing: development workflow and bug reports.
• Native packaging: release checks and corresponding source.

Linux developers can continue using run.sh after provisioning their environment; its usage is documented in the developer guide. All build and installation tools remain in this repository.

Logo pack

The complete 84-file logo pack includes outlined SVGs, transparent PNGs, light/dark and accent versions, Windows and macOS icons, favicons, social artwork, and a four-page usage guide. The desktop uses assets/branding/anharmonic-studios.svg for its icon and assets/branding/anharmonic-header.svg for the theme-aware project header. The pack's usage notes cover its formats.

Repository contents

• mpclab/: application and audio engine source.
• native/: C++ engine source and build configuration.
• assets/: application artwork and licensed factory instrument recordings.
• scripts/: build, installation, and validation tools.
• tests/: automated tests.
• website/: the public landing page, pricing, and one-minute commercial.
• packaging/: native release instructions and the public encryption certificate.
• BUILDING.md: platform-specific developer build instructions.
• CONTRIBUTING.md: source development and contribution guide.
• SUPPORT.md: support scope for official and self-compiled builds.
• DISTRIBUTION.md: free-source and paid-download model.

Personal libraries, songs, recordings, exports, development docs, agent automation, credentials, caches, and compiled builds are excluded. This repository starts with clean source history; it does not import the former development repository's history.

The orchestral recordings are CC0, with their dedication retained alongside the assets. Other dependencies and media keep their own licenses. See LICENSE and THIRD_PARTY.md.

Release candidates

Native 0.1.0-rc.1 candidates passed 227 regression tests per target, frozen-app checks, and platform packaging checks. The validated native build run covers Linux, Windows, Intel Mac, and Apple Silicon Mac. Physical audio-interface acceptance and corresponding-source review remain before production qualification. The release policy is unsigned distribution with clear installation guidance, not a future promise of signed or notarized packages.

Release CI uploads encrypted candidates, never plaintext paid installers. See the build guide and the distribution plan.