About RemoveVocals
Free AI audio tools built and operated by NOTHING SAS in Paris, France. Free to try, optional HD plans.
Our Mission
RemoveVocals is built for musicians, DJs, podcasters, educators, and content creators worldwide. Our mission is simple: make professional audio processing accessible to everyone, free to try, in 17 languages.
Privacy by policy
Vocal Remover and Stem Splitter separate your song on our GPU servers (Modal, United States) and send the result back to your browser; if the server declines, a 67 MB model downloads once and the separation runs on your device. Lyrics Finder sends the audio to our speech-to-text provider (Groq, United States) for the duration of the transcription. The editing tools (cutter, joiner, pitch changer, equalizer, bass booster, audio effects, converter, noise reducer, BPM finder, key finder, mastering, recorder) run entirely in your browser. Files sent to a server are deleted right after processing and never used to train models. Details in the privacy policy.
Our Technology
We run open source separation models (Open-Unmix and Demucs) on GPU servers for vocal and stem separation, with an in-browser WebAssembly fallback, and custom in-browser implementations for BPM detection, key finding, mastering and audio editing.
The Toolkit
15 specialized audio tools, each designed to do one thing exceptionally well.
Who We Are
RemoveVocals is built and operated by NOTHING SAS, a French company registered in Paris (SIREN 892 798 208), for the creative community. The 15 tools are free to use and we intend to keep them that way; HD downloads and 6-stem separation are part of the optional Basic, Pro and Studio plans.
Our Blog
We publish regular articles covering audio production techniques, tool comparisons, DJ guides, podcast tips, and more. Browse our blog to learn from our library of 52 articles.
Why the tools stay free to use
The editing tools run on your device, so each free use costs us close to nothing. Vocal and stem separation run on GPU servers that we pay for: the free tier gets a daily allowance on the standard model, and the paid plans fund the GPU time for HD separations.
The project funds itself through optional add-ons that do not hide any of the 15 tools: HD (GPU) export packs, the Basic, Pro and Studio plans (unlimited HD downloads, a library of your stems, 6-stem separation on Pro), and a one-time Founding Supporter pledge limited to 500 places. Eligible free pages display Google AdSense banners; paid members and Founding Supporters see no ads.
How the technology works
Vocal Remover and Stem Splitter upload your song over HTTPS to our separation server, which runs the model on a GPU and returns the stems to your browser; the temporary file is deleted right after processing. If the server declines the job (size, quota, load), a 67 MB model downloads once and the separation runs on your device with WebAssembly.
The editing tools decode the file with the Web Audio API, process it in memory in your browser and re-encode it to WAV or MP3 before the download starts. For those tools, no audio request leaves your device.
The company
RemoveVocals is published by NOTHING SAS, registered in Paris, France (58 rue de Monceau, 75008 Paris, SIREN 892 798 208). Contact: contact@removevocals.ai.
By the numbers
- 15 distinct audio tools shipped
- 17 languages for the site and the Lyrics Finder
- 0 copies kept of the audio sent for separation or transcription: files are deleted right after processing
- 0 watermarks added to exports
- 0 accounts needed to import, separate and listen; a free account only to download on Vocal Remover, Stem Splitter, Lyrics Finder and Mastering
- 52 articles in the editorial blog covering audio production, mixing, DJing and podcasting
- 500 Founding Supporter places, capped on purpose to keep the program meaningful
Press, partnerships and contact
For press inquiries, partnerships, sponsored content or integrations, see the partner page. For legal notices and entity details, see the legal notice. For privacy details, see the privacy policy.