StemPrep

StemPrep is a native macOS app that sends a WAV to MVSEP, tracks the separation job, and saves every returned stem beside the original track.

Highlights

• Live MVSEP model catalogue with the exact selected render_id
• Live queue position and real chunk progress inside the source waveform
• Model availability, credit estimates and supported advanced options
• WAV, FLAC and MP3 delivery options
• File-backed uploads avoid holding the full WAV in memory and remove their temporary staging file after submission
• Resumable job tracking, MVSEP history and finished-job re-downloads
• Queued-job cancellation through MVSEP when a refund is available
• Private account status for balance, premium use and active jobs
• Duplicate-render detection to avoid repeating paid jobs
• Output folders created beside the source track
• API token stored in macOS Keychain
• No analytics, advertising or developer-operated server

Requirements

• macOS 14 Sonoma or newer
• Intel or Apple silicon Mac
• An MVSEP account and API token
• A WAV source file

Install

1. Download StemPrep-v1.1.0.zip from the latest release.
2. Unzip it and move StemPrep.app to Applications.
3. Open StemPrep. If macOS asks for confirmation, right-click the app and choose Open.
4. Follow the first-run guide to create or sign in to an MVSEP account and save your API token.

StemPrep v1.1 is ad-hoc signed and is not notarized with an Apple Developer ID. The source is available here for inspection and local builds.

How it works

1. Drop a WAV into the app.
2. Choose the MVSEP separation model and output format.
3. Start the separation.
4. StemPrep writes a sibling folder named <Track Name> Stems containing the original, all returned stems and a JSON manifest.

Your audio is uploaded directly to MVSEP. StemPrep does not proxy it through another service. Read Privacy before using the app with sensitive material.

API token

Open the MVSEP User API page, create an account or sign in, then copy the API token shown on that page. StemPrep stores the token in macOS Keychain and uses it only for requests to MVSEP.

Build from source

Install Xcode 16 or newer, then run:

git clone https://github.com/lordydord/StemPrep.git
cd StemPrep
swift test
./script/build_and_run.sh --bundle

The universal app bundle is written to dist/StemPrep.app.

To install the local build in Applications:

./script/install_app.sh

Command-line analyser

The repository also contains the earlier Python audio-prep workflow. It estimates BPM and a first musical downbeat, creates a prepared WAV, and can submit it to MVSEP.

python3.11 -m venv .venv
.venv/bin/python -m pip install -r requirements.txt
cp .env.example .env
.venv/bin/python stem_prep.py process "/path/to/track.mp3"

The private .env file is ignored by Git. Never commit a real API token.

Project status

Version 1.1 adds truthful remote progress, account and credit context, advanced model options, MVSEP history, re-downloads and queued-job cancellation. Bug reports and focused pull requests are welcome through GitHub Issues.

StemPrep is an independent open-source project and is not affiliated with MVSEP or Ableton.

License

MIT