Remove the vocals from any song in one tap. Make karaoke tracks, pull clean acapellas, or hear just the drums or just the bass.

Stem Studio splits any song into four clean stems — Vocals, Drums, Bass, and Other instruments — right on your phone or tablet. Stem Studio runs the HT-Demucs v4 AI separation model 100% on-device. Your music is never uploaded, never analyzed in the cloud, and never leaves your device.

PRIVATE BY DESIGN
• 100% on-device processing: the AI model ships inside the app
• No internet permission at all: the app cannot send your audio anywhere
• No account, no sign-in, no tracking, no analytics, no ads

PAY ONCE, OWN IT FOREVER
• One purchase unlocks everything, for life
• No subscriptions, no credits, no per-song fees, no upload limits
• No server costs means no reason to ever charge you again

STUDIO-GRADE SEPARATION
• HT-Demucs v4 hybrid transformer: four real AI-predicted stems
• Isolate vocals for karaoke, covers, remixes, and transcription
• Pull out the drums or bass line to practice along with the band
• Mute any part to make backing tracks and play-alongs

BUILT-IN 4-STEM MIXER
• Synchronized waveform lanes for every stem
• Solo, mute, and per-stem volume faders
• A-B loop for practicing tricky passages
• Export each stem, a custom mix, or a ZIP of all four as WAV

WORKS WITH YOUR FILES
• MP3, M4A/AAC, FLAC, WAV, OGG, and the audio track of MP4 videos
• Songs up to 20 minutes
• Keeps working in the background while you switch apps

Separation is real AI work done on your phone's processor: a typical 4-minute song takes several minutes on a recent phone. Best on 64-bit devices with 4 GB of RAM or more.

Part of Revisited Labs: local-first creative and audio tools with zero subscriptions.