Recut

How to separate vocals from music in a song

Attach the song and ask Recut to "separate the vocals" or "split this into stems". It runs an audio separation model that splits the recording into the voice and the music (and, with some models, drums, bass and other parts).

What to typeSplit this song into vocals and instrumental.

Step by step

  1. Attach the songAn audio file, or a video: Recut takes the sound out first.
  2. Say what you needThe vocals on their own, the instrumental, or all the stems.
  3. Pick the modelRecut suggests separation models such as ElevenLabs audio isolation, Demucs and SAM Audio, with their prices.
  4. Run itEach part comes back as its own sound file on the timeline, ready to remix, mix or put on a video.

What Recut does

Separation models have learned what voices and instruments sound like, and split a mixed recording into parts. Demucs is a well-known model for music stems; voice-isolation models are tuned for pulling speech out of noise.

Once you have the parts, Recut can keep working with them on your device: change the speed, mix tracks together at different volumes, or put an instrumental under a new voiceover.

If you attach a video rather than a sound file, Recut first saves its sound as an MP3 on your device, then sends only that sound to the separation model.

Tips

What it doesn't do

Try it on your own video

Recut runs in your browser, on Mac, Windows and Android. Start free, attach a file and say what you want.

Open Recut in your browser

Related guides