AI Stem Splitter

The tool separates an audio file into six distinct stems—vocals, drums, bass, guitar, piano, and an “other” track—using the htdemucs model from Meta AI, which was a winner of the Sony Music Demixing Challenge. Users upload a track in common formats (MP3, WAV, FLAC, M4A) and receive the individual stems within about a minute, without needing a digital audio workstation or additional plugins.

It also offers a vocal‑removal mode that creates karaoke‑style instrumentals or isolates acapellas for remixing, with options to process audio directly from YouTube or SoundCloud URLs. The interface includes waveform previews, BPM and key detection, and export settings for DJ workflows such as Rekordbox.

The service operates on a credit system that never expires and provides an initial ten‑minute free quota, eliminating subscription requirements. It targets singers, DJs, remixers, and anyone needing quick, studio‑grade stem separation for practice, production, or content creation.