SampleBrain — Ableton Live Extension

Concatenative synthesis inside Ableton Live 12, inspired by SampleBrain by Dave Griffiths & Aphex Twin.

Made by Sernn

---

What it does

SampleBrain chops your brain tracks (audio corpus) into small grains, analyses each grain spectrally (MFCC or FFT), then reconstructs the target track selection by finding the closest-matching grain for each time slice. The result is placed on a new audio track.

Target (spectral shape) + Brain (sound material) → new audio clip

---

Install (no dev tools needed)

Requires Ableton Live 12 Beta, Suite edition.

1. Download SampleBrain-1.1.0.ablx from Releases
2. In Live: Preferences → Extensions
3. Drag & drop the .ablx file onto the panel

4. Make sure Developer Mode is OFF in Live's preferences
5. Right-click any audio track arrangement selection → SampleBrain…

Note: if the extension doesn't appear after install, try disabling other extensions temporarily — Live 12 Beta has a known bug where loading many extensions at once can break the catalog.

---

How to use

Step 1 — Select a time range on an audio track in the Arrangement View (click + drag)

Step 2 — Right-click → Extensions → SampleBrain…

Step 3 — Configure and Generate — check at least one brain source, adjust parameters, click Generate ▶

A new track SB: <target name> appears with the result.

Parameters

Parameter	Description
Block size	Grain size in ms. Small = granular. Large = smooth/tonal.
Overlap	OLA crossfade amount. 50% is standard.
Window	FFT windowing. Hann recommended.
Descriptor	MFCC = timbre match. FFT = spectral match. Mix = both.
Novelty	Bias against reusing grains. Higher = more variety.
Boredom	How fast novelty penalty resets.
Stickiness	Tendency to follow sequential grains. Higher = longer coherent runs.

---

Development setup (SDK step by step)

This section explains how to rebuild the extension from source using the Ableton Extensions SDK v1.0.0-beta.0.

Prerequisites

Tool	Version	Notes
Ableton Live 12 Beta	Suite	ableton.com
Node.js	≥ 24.14.1	nodejs.org
The Extensions SDK .zip	1.0.0-beta.0	Distributed by Ableton (Centercode)

Step 1 — Enable Developer Mode in Live

Open Live 12 Beta → Preferences → Extensions → enable Developer Mode.

Step 2 — Get the SDK

The SDK is distributed by Ableton via Centercode (beta program). The zip contains:

extensions-sdk-1/
├── ableton-extensions-sdk-1.0.0-beta.0.tgz
├── ableton-extensions-cli-1.0.0-beta.0.tgz
├── ableton-create-extension-1.0.0-beta.0.tgz
├── docs/
└── examples/

Copy the two .tgz files into vendor/:

cp path/to/ableton-extensions-sdk-1.0.0-beta.0.tgz vendor/
cp path/to/ableton-extensions-cli-1.0.0-beta.0.tgz vendor/

Step 3 — Install dependencies

npm install

Step 4 — Configure the Extension Host path

Create a .env file at the root of the project:

# macOS
EXTENSION_HOST_PATH=/Applications/Ableton Live 12 Beta.app/Contents/Helpers/ExtensionHost/ExtensionHostNodeModule.node

Note: The ExtensionHostNodeModule.node file is only present in the Beta build of Live, not in the release version yet.

Step 5 — Run in development mode

Important: Live must be open and fully loaded before running npm start.

# 1. Open Ableton Live 12 Beta
# 2. Then:
npm start

You should see:

Starting Extension Host...
  Extension: /path/to/samplebrain
  Live: /Applications/Ableton Live 12 Beta.app/...

info: #######################################
info: Started: Extension Host 1.0.0
info: #######################################
info: Extension Host sends greeting to Live
info: FlipMessageStreamSocket send success

The extension is now live. Right-click on an audio arrangement selection in Live to use it.

Step 6 — Build for production

npm run build

Step 7 — Package as .ablx

npm run package
# → SampleBrain-1.0.0.ablx

Drag the .ablx into Preferences → Extensions (Developer Mode OFF) to install it permanently.

---

Project structure

samplebrain/
├── src/
│   ├── extension.ts   # Entry point — context menu, HTTP server, main flow
│   ├── brain.ts       # SampleBrain algorithm — segmentation, kNN, OLA synthesis
│   ├── dsp.ts         # FFT (Cooley-Tukey), MFCC, windowing — pure TS, no deps
│   ├── wav.ts         # WAV + AIFF decode/encode — pure TS, no deps
│   ├── dialog.html    # UI — Ryoji Ikeda-inspired dot-matrix aesthetic
│   └── html.d.ts
├── vendor/            # SDK .tgz files (not on npm)
├── build.ts           # esbuild config
├── manifest.json      # Extension metadata
├── package.json
└── tsconfig.json

---

SDK key concepts

Execution model

Extensions are one-shot: triggered by right-click, run to completion, then stop. No persistent state, no real-time MIDI, no background loops.

export function activate(activation: ActivationContext) {
  const ctx = initialize(activation, "1.0.0");
  ctx.ui.registerContextMenuAction("AudioTrack.ArrangementSelection", "SampleBrain…", "cmd.id");
  ctx.commands.registerCommand("cmd.id", async (arg) => { /* ... */ });
}

Rendering audio

const wavPath = await ctx.resources.renderPreFxAudio(track, startBeat, endBeat);
// Returns a path to a WAV or AIFF file

Showing a dialog

// Serve your HTML over a local HTTP server, then:
const result = await ctx.ui.showModalDialog(`http://127.0.0.1:${port}/`, 576, 648);
// Resolves when the HTML calls close_and_send

Importing & placing a clip

const imported = await ctx.resources.importIntoProject(wavPath);
const newTrack = await song.createAudioTrack();
await newTrack.createAudioClip({ filePath: imported, startTime, duration, isWarped: false });

Progress dialog

await ctx.ui.withinProgressDialog("Processing…", {}, async (update, signal) => {
  update("Step 1…", 20);
  // ...
  update("Done.", 100);
});

---

License

MIT — do whatever you want with it.

---

SampleBrain algorithm originally by Dave Griffiths / then-try-this, developed with Aphex Twin.