Three simple steps
How to remove vocals from a song
- 01
Choose a supported audio file from your device. The song stays in your browser and is not uploaded to ToolYogi.
- 02
Click Remove vocals. Your browser uses WebGPU when available or WebAssembly fallback to separate the audio locally.
- 03
Preview the vocals and instrumental outputs, then download either WAV file. Refreshing or closing the page clears the session.
What does the Vocal Remover create?
The tool creates two outputs: a vocals stem and an instrumental stem. The instrumental is built from the non-vocal parts of the same local separation result.
Does ToolYogi upload or store my song?
No. The selected audio is decoded and processed in your browser. ToolYogi does not provide a cloud upload endpoint, file history or server storage for this tool.
Why can vocal removal take time?
AI separation is computationally intensive. Processing speed depends on your browser, memory and hardware. Devices with WebGPU support are generally faster than WebAssembly fallback.
Will every vocal be removed perfectly?
Not always. Reverb, backing vocals and instruments that overlap the human voice can leave artifacts or bleed between outputs. Results vary by recording and mix.
A little more clarity
Frequently asked questions
Is the Vocal Remover free?
Yes. The current ToolYogi Vocal Remover runs locally in your browser without requiring an account.
Does ToolYogi keep my audio?
No. Your song and generated outputs are not uploaded to or stored on ToolYogi servers. Temporary download URLs exist only in the current browser session.
What files do I get?
You get a vocals WAV and an instrumental WAV.
Does it use the same model as Stem Separator?
Yes. The current implementation uses the same local HTDemucs browser model, then combines the non-vocal stems into one instrumental output.
Do I need a powerful computer?
A modern device with WebGPU is recommended. The tool can attempt WebAssembly fallback, but processing may be much slower.
Can I process copyrighted music?
Only process audio that you own, have permission to use, or are otherwise legally entitled to process.
Open-source local processing
Vocal separation uses the same MIT-licensed demucs-web browser runtime and HTDemucs model path as ToolYogi’s Stem Separator. Audio processing stays on the user’s device.