Audio Convert Studio
๐Ÿ”’ Local processingโ˜๏ธ 0 file upload
SEPARATION ยท LOCAL AI

๐ŸŽค AI Vocal Remover

Remove vocals and create an instrumental version.

ACS EngineLoadingโ€ฆ
Backendโ€“
StatusLOCAL AI

Choose a file

Choose a music or video file

HTDemucs is a large model (~181 MB). For the first test, use 20โ€“30 seconds of music. WebGPU is much faster.
Waiting.
Your music stays in the browser. HTDemucs and ONNX Runtime run from AudioConvert hosting; no paid AI API is used.

Remove vocals from music with AI Vocal Remover

AI Vocal Remover separates the voice from a finished music recording and combines the drums, bass and other instrumental parts into an instrumental version. Use it to hear the accompaniment or practise singing over it. The result is a downloadable WAV audio file; your original file remains unchanged.

What can you use it for?

For example, create a rehearsal backing track from your own song. Listening without the lead vocal can help you study the arrangement or prepare a new vocal recording.

How to use it

  1. Choose music or a video containing audio.
  2. Start AI separation and wait for the model to load.
  3. Compare the original recording with the instrumental result.
  4. Use the WAV download button to save the accompaniment.

Tips for better results

Try a 20โ€“30 second excerpt first. A clean source is a better starting point than a distorted recording or audio that has been compressed repeatedly.

Limits and checking the output

Reverb, backing vocals and overlapping instruments can leave audible vocal remnants. Some instruments may also change. Separation estimates the components of a mix; it does not recover the original studio tracks.

Frequently asked questions

Will all vocals disappear?

Complete removal is not guaranteed. Reverberant vocals and backing voices can remain audible.

How is this different from Acapella Extractor?

This tool keeps the instrumental accompaniment. Acapella Extractor keeps the vocal part instead.

Processing and privacy

Your selected file is processed in your browser on your own device; it is not sent to an external AI processing service. Loading the required model and runtime for the first use can take time and consume data. Speed depends on recording length, browser support and device resources. On mobile, try a short excerpt first, keep the tab open and save the result. See the privacy notice for other data processing on the website. Privacy notice.

Related tools