Creative Tools / UNLOOP STUDIO
StemSplit
Vocals, drums, bass and more. Split your sound, entirely offline.
SEE IT IN ACTION
Try StemSplit.
Explore the controls below to see how it works. This is an interactive illustration; your files stay untouched.
A CLOSER LOOK
About StemSplit
Turn a song into parts you can work with. StemSplit separates vocals, drums, bass and other instruments on your computer, with its AI model and audio tools already bundled. No audio uploads, accounts or first-run model downloads.
Find the sound you need
Import MP3, WAV, FLAC, M4A, AAC or OGG. Choose Balanced mode for everyday separation or Studio mode for a more intensive pass. Adjust each stem’s volume, mute it or listen solo. Full mix, karaoke and a cappella presets help you get started.
Play, loop and refine
Hear the stems together with aligned playback. Seek using the waveform or timeline, set an A–B loop, and select a range to export. Reset A–B returns to the full track without losing your playback position. Your sessions stay in a local library.
Export your way
Save your current mix as 24-bit WAV, FLAC or 320 kbps MP3, or export all four original stems as individual 24-bit WAV files. Separation results depend on the source recording; processing speed depends on track length, mode and hardware.
Your audio stays with you
The standalone desktop app works offline, including its first launch. The model, Python runtime, Qt, PyTorch and FFmpeg are included, so there is no separate runtime or codec setup.
- Separate vocals, drums, bass and other instruments locally
- Balanced and Studio separation modes
- Per-stem volume, mute and solo with aligned playback
- Full mix, karaoke and a cappella presets
- Waveform seeking, A–B looping and range export
- Export mixes as 24-bit WAV, FLAC or 320 kbps MP3
- Export four individual 24-bit WAV stems
- Bundled AI model and audio dependencies; works offline from first launch
- Persistent local session library
Up and running.
macOS · Apple Silicon
Open StemSplit-1.2.1-macos-arm64.dmg and drag StemSplit into Applications. Open the app, import an audio file and select a separation mode. No separate Python, FFmpeg or model installation is needed.
Windows · x64
Run StemSplit-1.2.1-windows-x64-setup.exe and follow the installer. Launch StemSplit from the Start menu, import a song and start separation. The app includes its runtime and audio dependencies.
The supplied installers are not publisher-verified. Review your operating system’s installation prompts.
What’s changed
Version 1.2.1
Fixes a PyTorch normalization error when saving loud separated stems. Normalization creates a new tensor while preserving relative stem balance. Retains the light interface, bundled offline model and Reset A–B controls.
A few useful answers.
Is StemSplit free?
Yes. Open @unloopstudioofficial on Instagram, then return and confirm that you followed to unlock the free download. No payment or marketplace account is required. We use your confirmation and do not access your Instagram account or verify follower status.
Does my audio leave my computer?
Audio separation runs locally with the bundled model. The standalone app needs no cloud inference or audio upload, and works offline from its first launch.
Which stems can I separate?
Vocals, drums, bass and other instruments. Mix them with individual volume, mute and solo controls, or start from full mix, karaoke and a cappella presets.
Which export formats are available?
The current mix can be exported as 24-bit WAV, FLAC or 320 kbps MP3. All four unmodified stems can also be exported as individual 24-bit WAV files.
Will it perfectly isolate every sound?
Separation quality varies with the recording. Balanced and Studio modes offer different processing tradeoffs, and Studio mode takes longer.