Press
edytlab press kit
Everything in one place for writing about edytlab: a paragraph you can paste, facts that have been checked against the source, and the logo and demo files. There is no press office; questions go to GitHub.
Boilerplate
edytlab is a free, open-source desktop audio editor where you edit by chatting with an AI agent. The agent turns a request such as “time-stretch the second track to match the first, crossfade, and export a WAV” into calls to 93 deterministic audio tools. The tools run locally in a Rust engine on the user's own computer, so audio files are never uploaded; only the chat goes to the language-model provider the user selects (Anthropic, OpenAI, OpenRouter, Groq, Gemini, or Ollama for a fully local setup). Every edit is saved as a node in a branchable history, with undo and A/B comparison. edytlab runs on macOS, Windows and Linux, is released under the MIT license, and its source is at github.com/laadtushar/edytlab.
Fact sheet
| What it is | A desktop audio editor driven by an AI agent |
|---|---|
| Platforms | macOS (one universal build for Apple Silicon and Intel), Windows 10 and 11, Linux |
| Installers | .dmg, .msi and NSIS .exe, .deb and AppImage |
| Signing | Not signed or notarized yet, so macOS and Windows warn on first launch. The release notes give the steps. |
| License | MIT |
| Built with | Tauri 2, Rust, React |
| Audio engine | Pure-Rust DSP, including a phase vocoder written for the project for time-stretch and pitch-shift |
| Tools | 93 deterministic tools the agent can call (full reference) |
| AI providers | Anthropic, OpenAI, OpenRouter, Groq, Gemini, and Ollama (local, no key) |
| Where audio goes | Nowhere. Processing is local. Only chat text goes to the provider you choose. |
| Telemetry | None in the app; no analytics on this website |
| History | Branchable graph of edits with fork, A/B compare and undo |
| Safety option | “Plan first” shows the agent's steps for approval before they run |
| Export formats | WAV, FLAC, MP3 |
What is not shipped
Please state this plainly if you cover edytlab.
- Stem separation and speech transcription are not implemented in any build. Editing by transcript depends on transcription, so it does not work either.
- It edits files offline. It is not a live DJ tool and it is not a DAW: there is no MIDI and no VST or CLAP plugin hosting.
- There is no web or mobile version.
There are no user counts, download statistics, benchmarks, testimonials or awards to cite.
Logos and demo video
These files are in the project's MIT-licensed repository. There are no separate brand guidelines. Write the name in lowercase: edytlab.
- Logo (SVG) — Vector, for print and large sizes
- Logo (PNG) — 512 × 512
- Icon 512 — 512 × 512
- Icon 192 — 192 × 192
- Demo video (MP4) — Beatmatch and blend two tracks, 2:58, shown at 1.6× speed
- Demo video poster (JPG) — 1280 × 776
Screenshots of the interface are not published as files. The demo video shows the real app, and the release builds are free to download and capture yourself.
Contact
Open an issue at github.com/laadtushar/edytlab/issues. The maintainer is @laadtushar on GitHub.