Everything you need to know to get the most out of Stem Studio.
Stem Studio is a native macOS application that uses an on-device AI model to separate any song into its individual musical components — called stems. The entire process runs locally on your Mac using Apple's Neural Engine, so no audio is ever uploaded anywhere.
Whether you are a musician who wants to create a karaoke track, a producer studying an arrangement, or an audio engineer isolating individual instruments, Stem Studio gives you clean, high-quality stems in a few clicks.
The Stem Studio window is divided into two main areas:
Shows the active project with its status and metadata (format, duration, sample rate), followed by a searchable list of all saved projects. The AI model status indicator sits at the bottom.
Displays the drop zone when no project is active, or the stem track view with individual waveforms once separation is complete.
You can import audio in two ways:
Drag and drop — drag any audio file directly from Finder onto the centre drop zone. The file is loaded instantly and a new project is created automatically.
File menu — choose File → Open… (or press ⌘O) to pick a file from the standard macOS open panel.
Supported formats: MP3, WAV, AIFF and AAC (M4A). Stem Studio reads the file at its native sample rate — no re-encoding before processing.
Once you have loaded an audio file, the AI separation starts automatically. Here is what happens step by step:
The file is decoded and prepared for inference. The sidebar shows the project status changing to Separating…
Stem Studio uses Demucs v4 to process the audio in chunks on the Apple Neural Engine. A progress bar in the sidebar shows the overall completion percentage.
When inference is complete the six stem tracks appear in the centre panel, each with its own waveform visualisation.
Tip: Processing speed depends on the chip. On an M2 MacBook Air a 3-minute song typically takes 60–90 seconds. The app remains fully responsive during inference — you can manage other projects in the sidebar while waiting.
Stem Studio extracts the following tracks from every song:
| Stem | What it contains |
|---|---|
| Vocals | Lead and backing vocals, spoken word |
| Drums | Kick, snare, hi-hat, cymbals and all percussive elements |
| Bass | Bass guitar, synth bass and low-frequency tonal elements |
| Guitar | Electric and acoustic guitars |
| Piano | Piano, keyboards and most tonal keyboard instruments |
| Other | Everything that does not fall into the above categories: strings, brass, synth pads, sound design, etc. |
Note: Separation quality varies by recording. Tracks with many overlapping timbres (e.g. dense orchestral music) may have some bleed between stems. Stem Separator is optimised for contemporary pop, rock, electronic and acoustic music.
Each stem has its own row in the centre panel with independent controls:
Click the play button on any stem row to listen to that track in isolation. All stems are synchronised — scrubbing one scrubs all.
Toggle the mute button to silence an individual stem while the others continue playing — useful for A/B comparison.
Each stem has an independent volume slider so you can blend the tracks at custom levels during preview.
The waveform thumbnail gives a visual overview of the stem's energy over time. Click anywhere on it to jump to that position.
Every imported file creates a project that is automatically saved to disk. Projects store the original audio reference and all separated stems so you never need to re-process the same song.
Open a project — click any project row in the sidebar to load it into the centre panel.
Rename a project — hover over a project row and click the pencil icon, or hover the active project title in the upper sidebar. Press Return to confirm or Escape to cancel.
Delete a project — hover over a row and click the trash icon. You will be asked to confirm. This action permanently removes all stems and project data from your disk.
Search projects — use the search field at the top of the project list to filter by name.
Sort order — click the sort icon next to the Projects header to toggle between ascending and descending chronological order.
When separation is complete you can export all stems as individual WAV files:
Click the export icon (↑) in the active project header in the sidebar.
Choose a destination folder in the standard macOS save panel.
Stem Studio writes six WAV files — one per stem — to the chosen folder. A progress bar in the sidebar tracks the export.
When complete, a confirmation dialog appears with a Show in Finder button to open the export folder directly.
File naming: Exported files follow the pattern
ProjectName_stem.wav — for example MySong_vocals.wav,
MySong_drums.wav, etc.
| Requirement | Detail |
|---|---|
| macOS | 14 Sonoma or later |
| Chip | Apple Silicon (M1 or later) recommended. Intel Macs supported but slower. |
| RAM | 8 GB minimum; 16 GB recommended for large files |
| Storage | ~500 MB for the app and bundled model |
| Internet | Not required for any functionality |
How long does separation take?
On an M2 MacBook Air a 3-minute song takes approximately 60–90 seconds. Longer songs and older chips will take proportionally more time. The progress bar keeps you informed.
Can I use Stem Studio without an internet connection?
Yes, completely. All AI processing is on-device. Once the app is installed you need no internet connection at all.
My stems have some bleed from other instruments. Is that normal?
Stem separation is a hard problem, especially for dense mixes. Stem Studio's AI model is one of the best available, but perfect isolation is not always possible. Bleed is more common in heavily layered productions or where instruments share a similar frequency range.
What audio formats are supported for import?
MP3, WAV, AIFF and AAC (M4A) are supported. Other formats may work if macOS can decode them natively, but only the listed formats are officially tested.
Are my projects stored in iCloud?
Projects are saved in the app's local container on your Mac. If you have iCloud Drive Desktop & Documents syncing enabled in macOS, they may be included in your iCloud backup, but Stem Studio itself does not upload anything.
Can I run multiple separations at the same time?
Not simultaneously — the AI model processes one project at a time to ensure the best performance. You can queue files by creating multiple projects; each will be processed in turn.
The model status shows "Model unavailable". What should I do?
This means the bundled Core ML model could not be loaded. Try quitting and relaunching the app. If the issue persists, please contact support with the macOS version and Mac model you are using.
If you have a question not answered here, found a bug, or want to share feedback, please get in touch. We read every message and aim to reply within a few business days.
Send your questions or bug reports to support@stemstudio.app. Please include your macOS version and a brief description of the issue.