Ultimate Vocal Remover in Pro Tools

Hey - Yes this still works. I use it often to extract a voice over or music from a clip in Pro Tools. The current script is set up to make an instrumental and vocal of the selected file in Pro Tools. It is a workaround of a workaround, so the setup is a little complicated.
To start you need to install audio-separator on your computer. There are instructions on their site
https://pypi.org/project/audio-separator/.
For me this is done with a terminal command like 'pip install audio-separator'. Once it finishes, confirm in terminal with 'which audio-separator'. It will return a path like "/opt/homebrew/Caskroom/miniforge/base/bin/audio-separator" You will need this path in your script - Here is the updated script I use, you'll need to change a few things specific to where audio-separator gets installed on your computer:

sf.ui.proTools.appActivateMainWindow();

//Selection
const selection = sf.ui.proTools.selectionGet();
const timecodeIn = `${selection.selectionStart}`;

// Set Pro Tools session to interleaved
sf.app.proTools.setSessionInterleavedState({ interleavedState: true });

// Consolidate the selected clip in Pro Tools
sf.ui.proTools.menuClick({ menuPath: ['Edit', 'Consolidate Clip'] });
sf.ui.proTools.waitForNoModals();

sf.app.proTools.invalidate();

// Get the path of the selected file in Pro Tools
var fullPath = sf.app.proTools.getFileLocation({
	fileFilters: ['SelectedClipsTimeline'],
}).fileLocations[0].path;


// Build paths
var directoryPath = fullPath.substring(0, fullPath.lastIndexOf('/'));
var fileName = fullPath.substring(fullPath.lastIndexOf('/') + 1, fullPath.length - 4);
fileName = fileName.replace(/\.L|\.R$/, '');

// Disable interleaved again
sf.app.proTools.setSessionInterleavedState({ interleavedState: false });


// Output files expected by UVR naming
var vocalFile = `${directoryPath}/${fileName}_(Vocals)_UVR-MDX-NET-Inst_HQ_4.wav`;
var instrumentalFile = `${directoryPath}/${fileName}_(Instrumental)_UVR-MDX-NET-Inst_HQ_4.wav`;

// command line
sf.system.exec({
    commandLine: `PATH="/opt/homebrew/bin:/opt/homebrew/sbin:$PATH"; \
/opt/homebrew/Caskroom/miniforge/base/bin/audio-separator \
"${fullPath}" \
--model_filename UVR-MDX-NET-Inst_HQ_4.onnx \
--output_dir "${directoryPath}" \
--output_format WAV`
});

//Destination Tracks
 sf.app.proTools.selectTracksByName({
        trackNames: ["Dial 7"],
    });

// Copy the file to the clipboard
sf.file.copyToClipboard({ path: vocalFile });
sf.ui.proTools.spotFileFromClipboard();
sf.ui.proTools.waitForNoModals();

 sf.app.proTools.selectTracksByName({
        trackNames: ["Music_10"],
    });

// Copy the file to the clipboard
sf.file.copyToClipboard({ path: instrumentalFile });
sf.ui.proTools.spotFileFromClipboard();
sf.ui.proTools.waitForNoModals();    

Now in this section you need to change the script to work with your install location:

// command line
sf.system.exec({
    commandLine: `PATH="/opt/homebrew/bin:/opt/homebrew/sbin:$PATH"; \
/opt/homebrew/Caskroom/miniforge/base/bin/audio-separator \
"${fullPath}" \
--model_filename UVR-MDX-NET-Inst_HQ_4.onnx \
--output_dir "${directoryPath}" \
--output_format WAV`
});

You need to swap my install path "/opt/homebrew/Caskroom/miniforge/base/bin/audio-separator " with the install location that was returned when you entered 'which audio-separator' in terminal.

I have this setup up to work with my template, so you will need to change the names of these destination tracks to something that works in your session, in my case here the tracks are named Dial 7 and Music_10, these both need to be stereo tracks!

//Destination Tracks
 sf.app.proTools.selectTracksByName({
        trackNames: ["Dial 7"],
    });

// Copy the file to the clipboard
sf.file.copyToClipboard({ path: vocalFile });
sf.ui.proTools.spotFileFromClipboard();
sf.ui.proTools.waitForNoModals();

 sf.app.proTools.selectTracksByName({
        trackNames: ["Music_10"],
    });

Try directly using the updated python version in your terminal command
python3 -m pip install "audio-separator[cpu]"
or
pip3 install "audio-separator[cpu]"

When I first set this up I was using an M1, so should be all good

The problem is I'm not a python or terminal expert at all ! So I dont know if a video would actually be very helpful or accurate for everybody. I have set this up on a few computers, and it's always a little specific to that computer and python setup. On some I've had to create a virtual environment and install python/audio-sep fresh in that location. On others it was much easier by following the terminal steps I outlined above. I can look at expanding the script to include all the models and making this more user friendly
That said I think whats going on in your case is a mismatch between a newer version of python and an older version of onnxruntime, but again I'm not a python expert !