Isolate vocals, drums, bass, guitar, piano, and more. Practice at your own pace — 100% offline on your iPhone.
Free to try. One-time purchase for unlimited access.
Powered by state-of-the-art AI — running entirely on your device.
Split any song into vocals, drums, bass, guitar, piano, and other instruments using on-device machine learning.
Slow down playback, shift pitch, set A/B loops, and use the built-in metronome to learn at your pace.
All processing happens on your iPhone. No uploads, no accounts, no servers. Your music never leaves your device.
Automatic tempo detection with a synced metronome. Set count-ins for loop restarts to stay in time.
Export individual stems or custom mixes as WAV (lossless) or M4A (compressed) for use in your DAW.
One-tap practice modes: Karaoke, Guitar Practice, Drumless, Bass Practice, Vocals Only, and more.
Import any song from your Music Library or Files app.
AI splits it into 6 individual tracks in the background — listen to the original while stems prepare.
Solo, mute, slow down, loop, and play along.
Stem separation (also called source separation) uses AI to split a mixed song into its individual instrument tracks — such as vocals, drums, bass, guitar, piano, and other instruments. This lets musicians isolate or mute specific parts for practice or study.
No. All processing happens entirely on your iPhone or iPad using on-device machine learning. Your music never leaves your device and no internet connection is required.
Deconpose separates songs into 6 stems: vocals, drums, bass, guitar, piano, and other instruments. You can solo or mute any combination to create the perfect practice track.
Processing runs entirely on your device — no servers, no queues. Powered by Apple-silicon-accelerated on-device AI, a full song is typically ready within a few minutes on a recent iPhone — the latest models separate even a 7-minute song in under a minute, while staying easy on your battery. This is also what makes the one-time purchase model possible: there are no cloud servers to pay for month after month, so neither do you. While stems prepare in the background, you're not left waiting: hit play on the original track, slow it down, shift the pitch to your instrument's key, or set up your loops. All practice tools are available immediately — stems arrive while you explore the song.
Deconpose uses Demucs v4, one of the most advanced AI models for source separation. For most pop, rock, folk, and electronic music you'll get clean, practice-ready stems. Densely layered productions or heavily processed sounds may occasionally show some bleed between tracks — no AI model separates every recording perfectly. In practice, even imperfect stems are more than good enough for learning a part, slowing things down, or singing along — which is exactly what Deconpose is designed for.
Your first song is free — a complete separation with no time limit, and the stems stay yours with unlimited playback. A one-time premium purchase unlocks unlimited songs and stem export. No subscription required.
No. Deconpose collects zero personal data. There are no accounts, no analytics, no trackers, and no servers. Your audio files and usage stay entirely on your device. Read our Privacy Policy for full details.
Deconpose supports any audio format that iOS can read, including MP3, AAC, WAV, ALAC, FLAC, AIFF, and more.
Yes. With the premium upgrade, you can export individual stems or custom mixes as lossless WAV or compressed M4A files for use in any DAW or audio application.
Deconpose requires iOS 26 or later — that means iPhone 11 and newer, and any iPad that supports iPadOS 26.
Unlike online services, Deconpose processes audio entirely on your device — no uploads, no waiting for server queues, and complete privacy. It also includes built-in practice tools like speed control, pitch shifting, A/B loops, and a metronome that online tools don't offer.
Questions? support@deconpose.com