About AudioSplitAI

AudioSplitAI is a free online tool that separates a song into vocals, drums, bass and other instruments. The AI runs on your own device, inside your browser.

Why it exists

Many online vocal removers ask you to upload your music to a server, wait in a queue, or pay. We wanted a simpler option: open a page, choose a song and let your own computer do the work. Your files stay with you.

What you can do

• Split a song into four stems: vocals, drums, bass and other.
• Create a karaoke version in one tap.
• Preview and mix the stems with volume, mute and solo controls.
• Download WAV files, or a ZIP when you export several tracks.

How it works

The site uses a Hybrid Transformer Demucs model converted to the ONNX format. It is executed in your browser with ONNX Runtime Web. The model is downloaded once from Hugging Face and stored in your browser, and the separation itself happens on your device. For a plain-language explanation, read how AI audio separation works.

Limits we are honest about

• The stems are an AI estimate. Small traces of other instruments can remain, especially in dense mixes.
• Processing speed depends on your device. Computers are usually faster than phones, and long songs need more memory.
• Use only music you own or have permission to use.

The Android app

The same idea is available as an Android app, Audio Splitter Pro, with MP3 export, voice recording over your music and a library of your songs. Get it on Google Play.

Funding and advertising

The website is free. It may show advertising on its information pages to cover small running costs. The splitter itself does not show ads. Details are in the Privacy Policy.

Contact

Questions, bugs or ideas? See the contact page.