Free audio tools that respect your privacy
We build simple, honest tools for people who work with songs. No accounts, no uploads and no watermarks.
Why we built this
Most online vocal removers ask you to upload your song, create an account and wait in a queue. That is a lot of friction for a karaoke track, and it means handing your files to someone else.
We wanted a tool that works the moment the page loads, keeps your audio on your own device and is free for everyone to use. Singers, DJs, teachers and video creators all use it in different ways, and every one of them gets the same private engine.
What we promise
- Your files are never uploaded
- No account or email to use the tools
- No watermark on anything you download
- Clear answers about what works and what does not
How the engine works
A short tour of what happens in your browser when you drop in a song.
Reads your file
Your browser decodes the audio into raw sound, slice by slice, without uploading anything.
Recognises the voice
An AI model trained on thousands of songs predicts which parts of the sound belong to the voice and which belong to the music.
Keeps the band intact
Drums, bass and other instruments are separated too, then combined into a clean instrumental.
Builds two files
The voice becomes one track and the music the other, both saved as full quality WAV files.
Honest about the limits
The AI engine runs HTDemucs, an open-source model by Meta, entirely on your device, so it is private but depends on your hardware. It does a great job on most recordings and struggles in a few cases, and we would rather tell you that up front.
- AI separation works best on a device with a graphics card
- Live recordings and heavy reverb leave faint traces
- The first AI run downloads the model once, about 180 MB
- Better source files always give better results
More free audio tools
Every tool runs on the same private, in-browser engine. Pick the one that matches your job.
Try it on your favourite song
Free, private and ready in about a minute.