AI Stem Splitter

Tentang AI Stem Splitter

The AI Stem Splitter takes a finished song apart and hands back its four separate layers — vocals, drums, bass and everything else — using the open-source Demucs neural network. Every run separates the full mix, so you never have to decide up front which part you need.

Each stem comes back as its own MP3 that you can play directly on the page and download on its own, and all four are also bundled into a single ZIP. Two quality modes are offered: the standard htdemucs model for speed, or the fine-tuned htdemucs_ft model for a cleaner split.

Separation runs on the server and uses credits from your account. A deep neural network analyses the entire track, so a full-length song takes minutes rather than seconds — keep the page open while it works. Uploads and results are cleaned up automatically afterwards.

Cara menggunakan AI Stem Splitter

1. Upload the song you want to take apart — MP3, WAV, M4A or OGG.
2. Pick the quality: standard (htdemucs) for speed, or high quality (htdemucs_ft) for the cleanest split.
3. Play the original in the preview player to confirm you loaded the right track.
4. Click “Split into stems” and wait — a full song can take a few minutes.
5. Play each stem, download the ones you need, or take all four at once as a ZIP.

Pertanyaan yang sering diajukan