A lightweight, cross-platform audio editor and recorder — built in C++ on JUCE and released under GPLv3 — lets Claude and other MCP agents inspect and edit a session conversationally while it stays open on screen.
2026-10-02 · by Moe Ameen
Audionaut, a free multitrack audio editor and recorder for macOS, Windows, and Linux, is open source. Its creator, who goes by kvoltmer, released the code base under the GPLv3 license on August 15, 2026. The project's dual-license model also names a commercial license, but its own license notes say that option isn't actually available yet, pending agreements for the GPL/AGPL-linked third-party libraries it uses. The app is built in modern C++ on the JUCE framework and has shipped steady releases since, recent ones adding MP3 (via LAME), FLAC, and Ogg Vorbis export on top of WAV and AIFF.
What makes it notable is less the editor than who can operate it. Audionaut speaks the Model Context Protocol (MCP), so Claude and other MCP-capable agents can inspect and edit a project conversationally while it stays open in front of you. You add it with a single command — `claude mcp add audionaut -- npx -y audionaut-mcp` on Node.js 18+ — and the server exposes roughly two dozen tools covering project management, splitting and moving clips, adjusting gain, speed, and fades, naming regions, running beat detection and segmentation, separating stems, and assembling arrangements. Each agent edit arrives as a single undo step, and the UI shows exactly what changed.
Stem separation runs entirely on your machine through demucs.cpp, splitting a clip into drums, bass, vocals, and other without the audio leaving your computer. An Auto Edit feature uses on-device musical analysis — beat tracking, onset detection, and segmentation — to cut clips, and the core workflow stays close to the old regions-to-playlist-to-export model: mark regions, drop them to a playlist, export. Time-stretch spans roughly ×0.25 to ×4, and projects save in an open JSON format.
The honest framing: Audionaut is a production tool, not a content or distribution platform. It records, edits, analyzes, and exports audio — and stops there. It writes no captions, generates no video, and posts nothing to a feed.
Audionaut closes the production gap; the gap it leaves wide open is distribution, and that is where [Kompozy](/) picks up. Picture recording a video podcast and cleaning the audio in Audionaut, maybe with Claude driving the repetitive edits over MCP. The finished episode is one asset — but a week of content is the actual deliverable, and an editor can't produce it. Point Kompozy at the episode and it fans the ideas into the promo set an audio tool never touches: [Clipped Shorts](/glossary/clipped-short) cut vertical highlights from the long recording, [Persona Shorts](/glossary/persona-shorts) and HeyGen avatar video build a talking-head trailer, brand-exact [Carousel Posts](/glossary/hyperframes) walk through the sharpest quotes, and the transcript becomes a Blog Article of show notes plus an Email Newsletter.
A [Persona Brief](/glossary/persona-brief) keeps every output in one voice with your banned words enforced, and [Autopilot](/glossary/autopilot) with a per-post review step schedules the batch across eight social platforms plus blog and email from one queue. The split is clean: Audionaut — optionally with an agent — produces the audio; Kompozy produces and publishes everything that surrounds it. See [content repurposing](/glossary/content-repurposing) for the fuller one-source-to-many-outputs workflow.
Audionaut is a free, open-source multitrack audio editor and recorder for macOS, Windows, and Linux, built in modern C++ on the JUCE framework. It handles recording, cutting, regions and playlists, multi-channel editing, time-stretch, local stem separation, and export to WAV, AIFF, MP3, FLAC, and Ogg Vorbis — and it can be driven by AI agents over MCP.
Yes. Its code base went open source on August 15, 2026 under the GPLv3 license. A commercial license is named in the project's dual-license model but, per its own license notes, isn't actually available yet. It's free to download and use from the project's GitHub and site across all three desktop platforms.
Audionaut ships an MCP server. With the app open and Node.js 18+ installed, you run `claude mcp add audionaut -- npx -y audionaut-mcp`, and Claude or another MCP-capable agent can create projects, split and move clips, adjust gain, speed, and fades, name regions, run beat detection and segmentation, separate stems, and assemble arrangements. Each edit lands as a single undo step in the live UI.
No. Audionaut records, edits, analyzes, and exports audio — it writes no captions, makes no video, and publishes nothing to social platforms. To turn a finished episode or track into clips, short-form video, a carousel, show notes, and scheduled posts, pair it with a content engine like Kompozy.