AI Stem Separation

Isolate vocals, drums, bass, and melody from any track in real-time. Create acapellas, build live mashups, and apply effects to individual parts of a song.

The Four Stems

Tremor's AI engine separates every track into four distinct stems. This processing happens entirely on your device using the Neural Engine.

🎤
VOCALS
Singing, rap, spoken word
🥁
DRUMS
Kicks, snares, hats, percussion
🎸
BASS
Sub bass, bass guitar, 808
🎹
MELODY
Synths, piano, guitar, pads

How to Use Stems

Accessing the Stems Page

On iPhone, swipe to the STEM page in the center panel. On iPad, the stems controls are in the deck's expanded pad view. Each stem has three controls:

Starting Separation

Stem separation begins automatically when you load a track. The first pass takes a few seconds on Apple Silicon devices. Results are cached, so the same track loads instantly next time.

Creative Applications

🎤

Acapella Transitions

Mute the drums and bass on your outgoing track, leaving just the vocals floating over your incoming beat. A classic technique for smooth, musical transitions.

🥁

Instrumental Drops

Cut the vocal stem right before a drop to build tension. The crowd knows the words — let them sing along while the beat rides.

🎵

Live Mashups

Play the vocals from Track A over the drums and bass of Track B. No pre-production required — create mashups on the fly during your set.

Stem-Specific FX

Apply reverb only to vocals. Put a filter sweep on just the melody. Add delay to drums. With stem FX routing, your creativity has no limits.

Performance Tips

💡

Pre-Analyze Your Library

Before a gig, enable pre-analysis in Settings. Tremor will process stems for your entire library in the background, so tracks load instantly during your set.

Best on Apple Silicon

Stem separation uses the Neural Engine on A14+ iPhones and M-series iPads. Processing is virtually instant. On older devices, CPU fallback is used with slightly higher latency.