ablx.directory

RMX Toolbox

v0.0.1

Download audio from YouTube and separate them using audio-separator directly in Ableton (Extensions SDK).

2wolmanifest author: TheWolfNo ratings yet
README · rendered from GitHubView source

Ableton YouTube Separator

Download audio from YouTube and separate them using audio-separator directly in Ableton (Extensions SDK).

Quickies

Troubleshoting:

(Warn: Using CPU)

Not interesting:

(Why pipx)

Requirements

Ableton Suite 12.4.5 or newer

Node.js if you want to build the extension by yourself

yt-dlp

FFmpeg

Python 3.11

Installation

Windows

  1. Install Python 3.11 & pipx
winget install Python.Python.3.11
py -3.11 -m pip install pipx
py -3.11 -m pipx ensurepath

# now restart PowerShell
  1. Install yt-dlp & FFmpeg
winget install yt-dlp.yt-dlp
winget install Gyan.FFmpeg
  1. Install audio-separator
pipx install "audio-separator[gpu]" (or [cpu] for non-nvidia gpu cards)
  1. Verify
yt-dlp --version
ffmpeg -version
audio-separator --help

MacOS

  1. Install Homebrew
  2. Install Python 3.11, pipx, yt-dlp & FFmpeg
brew install python@3.11 pipx yt-dlp ffmpeg
pipx ensurepath
  1. Install audio-separator
pipx install "audio-separator[gpu]" (or [cpu] on older MacBooks with intel)
  1. Verify
yt-dlp --version
ffmpeg -version
audio-separator --help

Building

  1. Clone Repository
git clone https://github.com/2wol/Ableton-YouTube-Separator.git
  1. Open cloned repository in Visual Studio Code, open terminal and run this command to install missing packages
npm install
  1. Create .env file in the root directory and specify Extension Host Node Module path in it
# Example:
EXTENSION_HOST_PATH=C:\ProgramData\Ableton\Live 12 Beta\Program\ExtensionHost\ExtensionHostNodeModule.node
  1. Start for debugging or package minified build
npm start (or) npm run package

In progress dialog there is a message saying "(Warn: Using CPU)"

If you have an Nvidia GPU, here is the fix:

First, replace audio-separator Torch 2.12.0 with Torch 2.11.0+cu128 and check if the message pop up again.

pipx runpip audio-separator install --force-reinstall torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

If the message is still there, install CUDA Toolkit 12.8.0

If it's still there, then idk you cooked bro it works on my machine

Some not interesting stuff

why pipx

i don't know, I used that to not worry about "finding" FFmpeg and audio-separator separatelly on Windows and MacOS.

why

doing all that (separating vocal, DeReverbing it) manually in UVR5 was pain in ahh, now everything is under few click thanks to papa ableton.

Rate this extension

One rating per account; you can change it any time.

No ratings yet — be the first

Comments

Sign in to join the discussion.Sign in

No comments yet.