Professional vocal & instrumental separation,
entirely on your Mac.
Extract clean vocals and instrumentals from any track — without ever touching the internet. Just drag in a file, and the AI gives you two beautifully separated stems.
One-time purchase · £9.99
Truly Sandboxed
Runs inside Apple’s own security sandbox. It never asks for internet access.
Real AI, Locally
Uses the HTDemucs architecture, running right on your machine with zero latency.
Your Audio, Your Business
No accounts, no telemetry. The AI processes everything locally, and the stems are yours.
Deep learning, made simple.
We’ve packaged a full Python 3.12 environment, PyTorch, and the official HTDemucs model into a single app. There’s no terminal, no setup — just open DeepStem Pro and start separating stems.
The model processes stereo audio natively, preserving the original warmth and depth up to 48 kHz.
Python 3.12
PyTorch
HTDemucs
SwiftUI
Hear the difference.
The vocals sound full and present, the instrumental is rich and intact. No artifacts, no hollow sounds.
Vocals
Instrumental
Multitrack Studio Edition is coming.
We’re building a dedicated Studio Edition that will split drums, bass, piano, and more. It will be released as a separate standalone app.
Get DeepStem Pro
One-time purchase. No subscriptions. Yours forever.
Secure purchase via Apple