Text to SFX

An Ableton Live extension that generates a sound effect from a text prompt using the ElevenLabs Sound Generation API and drops it straight into a Session-view clip slot.

Right-click any clip slot → Generate SFX… → describe the sound, set a duration, hit Generate. The extension calls ElevenLabs, imports the result into your project, and creates an audio clip named after your prompt — all in one undo-friendly step.

Download

⬇ Download Text-to-SFX-1.0.0.ablx

Drop the .ablx onto Live's Preferences → Extensions page (Ableton Live 12 Beta with Extensions enabled). On first Generate SFX… you'll be prompted for your own ElevenLabs API key, which is stored locally for next time. Developer Mode is not required to use an installed extension.

How it works

1. A modal webview collects a prompt and a duration (0.5–30 s).
2. POST https://api.elevenlabs.io/v1/sound-generation returns MP3 audio.
3. The MP3 is written to the extension's temp directory.
4. resources.importIntoProject() copies it into the Live project.
5. clipSlot.createAudioClip() places it in the right-clicked slot.

Steps 2–5 run inside a native progress dialog with a working Cancel button.

Prerequisites

• Node.js ≥ 24
• Ableton Live Beta with Extensions support, Developer Mode enabled (Preferences → Extensions → Developer Mode).
• An ElevenLabs API key (https://elevenlabs.io/app/settings/api-keys). Each user supplies their own — the extension prompts for it on first use and stores it locally.

Setup

git clone https://github.com/serafinsanchez/elevenlabs-txt-to-sfx.git
cd elevenlabs-txt-to-sfx
npm install   # installs the vendored Ableton Extensions SDK + CLI from ./vendor
# Optional (dev convenience): put your key in .env so you aren't prompted while developing.
cp .env.example .env   # then edit .env and paste ELEVENLABS_API_KEY

The Ableton Extensions SDK is currently a private beta and is not published to npm, so the @ableton-extensions/sdk and @ableton-extensions/cli packages are vendored as tarballs under vendor/ and referenced via file: paths in package.json.

The API key resolves in this order: a key you've already saved in the extension (storageDirectory/config.json) → the ELEVENLABS_API_KEY env var (loaded from .env in dev) → an in-app prompt. Installed users have no .env, so they enter the key in the prompt and it's saved for next time.

Run

The CLI needs the path to your Live install (the Extension Host). --temp-directory guarantees the extension has a writable temp dir for the downloaded audio.

# macOS
npm start -- --live "/Applications/Ableton Live 12.4 Beta.app" --temp-directory ./.tmp

# Windows
npm start -- --live "C:\ProgramData\Ableton\Live Beta\Program\Ableton Live Beta.exe" --temp-directory ./.tmp

Then in Live's Session view, right-click a clip slot → Generate SFX….

Package for distribution

npm run package

This runs a production build and produces Text-to-SFX-<version>.ablx. Install it by dropping it onto Live's Preferences → Extensions page. The bundle is fully self-contained — both webview modals and all code are inlined, and Live provides the Node runtime — so end users only need Live Beta with Extensions enabled (Developer Mode is not required to use an installed extension). On first Generate SFX…, the extension prompts each user for their own ElevenLabs API key and stores it locally.

Troubleshooting

Logs go to your terminal (stdio is inherited) and to the Extension Host log: ~/Library/Preferences/Ableton/Live 12.x.x/ExtensionHost.txt (macOS).

• 401 — invalid API key. The extension clears the stored key and re-prompts; paste a valid key from https://elevenlabs.io/app/settings/api-keys.
• 422 — duration outside the 0.5–30 s range.
• "no temp directory available" — add --temp-directory ./.tmp to the run command.
• No menu item — confirm Developer Mode is on and the host stayed connected.

Notes

• model_id is omitted so ElevenLabs uses eleven_text_to_sound_v2, which is required for duration_seconds to take effect.
• Output is MP3 (mp3_44100_128); Live imports it directly. Switch the output_format query param to a pcm_* value (and add a WAV header) if you need lossless audio.