7 Tips for the Cleanest Possible Vocal Separation

AI separation is good out of the box, but a few habits will get you noticeably cleaner results. Here are the ones that matter most.

  1. Use the best source file you have. Quality in, quality out. A WAV or a high-bitrate MP3 beats a tiny low-quality file every time.
  2. Prefer studio recordings. Clean studio mixes separate better than live recordings full of crowd noise and room echo.
  3. Use Chrome on a computer for tough tracks. More cores and a graphics chip mean faster, more reliable processing, especially for long songs.
  4. Let the first download finish. The AI model downloads once. Don’t close the tab mid-download or you’ll start over.
  5. Pick the right mode. Just need a karaoke track? Use Vocals & Instrumental. Want drums and bass too? Use the full Splitter — but keep it to shorter songs on weaker devices.
  6. Listen on headphones. Small artefacts are easier to spot on headphones, so you know whether a result is good enough for your purpose.
  7. Export WAV for editing, MP3 for sharing. If you’ll process the audio further, keep the full-quality WAV; if you’re just sending it to a friend, MP3 is smaller.

None of these are complicated, but together they’re the difference between an okay result and one that sounds genuinely clean.

Try the free Vocal Studio tool now →