Karaoke maker

Upload a song and get it back without the singing. You get the instrumental as a karaoke track, and the vocal on its own, both as WAV files. Hear a 21-second preview of the separation before you process the full song.

This page needs JavaScript for the tool itself. Open it in a browser to use it.

How it works

Upload an MP3, WAV, FLAC or AIFF file. The tool separates one energetic section of the song and plays you the instrumental and the vocal so you can judge the result. If you like it, process the whole song; the karaoke track appears in your dashboard when it is ready.

A karaoke track that keeps the song

Older karaoke methods cancelled the centre of the stereo mix and took the bass and drums with it. The AI model listens to the whole mix and takes out only the voice, so the instruments, the reverb and the stereo picture stay as they were recorded.

Practise, perform, record

Sing over the instrumental to learn a song, put it under your own lyric video, or record a cover on top of it. The isolated vocal that comes with it is useful too: study the phrasing of the original singer with the band taken away.

How to make a female version cover

Vocal remover, stem splitter, same engine

This page, the vocal remover and the AI stem splitter all run on one separation model. Choose 4 stems on the preview screen if you also want drums, bass and the other instruments as separate files.

Open the AI stem splitter

Made by a label of singers

BeatRootRaga is an independent label that records ghazals, Indian classical and film-song covers. The karaoke maker began as the label's own practice tool for singers learning a song from a reference recording.

Featured releases

All releases

Latest guides

All guides

Questions people ask

Is the karaoke maker free?

Yes. Upload a song and hear a 21-second preview of the karaoke track without signing up. An account is needed to process the full song and download it, so the file can be kept in your dashboard.

How does it make a karaoke track from a song?

An AI model (Demucs, an open-source hybrid transformer for music source separation) separates the singing from the instruments. The instrumental is your karaoke track; the isolated vocal comes with it as a separate file.

Will the vocals be gone completely?

Usually close to it. A clean studio vocal separates well; a dense mix with heavy reverb or doubled voices keeps faint traces. The preview lets you judge the result before you process the whole song.

Does it show lyrics or make a video?

No. It produces the audio only: a stereo WAV of the instrumental, and the vocal on its own. Play the WAV in any player or put it under your own lyric video.

Which songs can I use?

Any song file you upload, MP3, WAV, FLAC or AIFF up to 100 MB. Making a karaoke track does not change who owns the recording, so use it for songs you own, have permission to use, or that are licensed for your purpose.

Can I keep the backing vocals?

The model separates all voices from the instruments, lead and backing together. If you want harmonies kept, the 4-stem split still puts every voice in the vocal file, so that is not possible here.