Remove vocals from any song with AI. Free, no sign-up, 3 songs a day — uploads auto-deleted in 24h. Create instrumentals and karaoke tracks instantly.
State-of-the-art source separation — the same class of models used in professional DAWs.
Files are uploaded for processing and deleted immediately after. Never stored or shared.
3 free jobs per day, files up to 20 MB. No account required.
Upload your song — MP3, WAV, FLAC, OGG, M4A or AAC, up to 20MB.
The AI separates vocals from the instrumental automatically (usually under a minute).
Preview both tracks, then download the instrumental or the isolated vocals as MP3.
Upload the track, wait a few seconds, and download the instrumental. That's the whole workflow — no account, no software install, no watermark. The free tier covers 3 songs per day, which is enough for karaoke nights, DJ edits, and remix projects.
Old-school vocal removers used phase cancellation: because lead vocals sit in the stereo center, subtracting the left channel from the right cancels them. It also destroys the bass, kick, and snare — anything else panned centrally — and fails completely on mono files.
Modern tools use a deep neural network called Demucs, trained on thousands of songs to recognize what vocals sound like versus instruments. The model listens to the mix and reconstructs two separate sources: a clean vocal track and a full instrumental that keeps the bass and drums intact. That's the same technology behind professional software like Serato Stems and Native Instruments stems.
Most 'free' vocal removers trade you something: a sign-up wall, watermarks, 30-second previews, or a paid download step. This one has none of that — no sign-up, no watermark, just upload, listen, and download the full MP3. The daily limit exists only to keep the service free; every separation genuinely costs compute, so 3 per day keeps it sustainable without charging anyone.
Yes — this one. No account, no email, no software to install: open the page, upload a song, and download the separated tracks as MP3. The free tier is 3 songs per day with no watermark on the output. It's the honest kind of free: the daily cap exists because every separation runs a real AI model, not as an upsell trick.
The separation is very good on typical studio mixes — the bass and drums stay intact. Dense mixes with heavy reverb on the vocals or live recordings can leave faint vocal ghosting. Preview both tracks before downloading to judge for yourself.
Yes. Because the AI recognizes vocals rather than relying on stereo phase tricks, mono files and vintage recordings separate as well as modern stereo mixes.
MP3, WAV, FLAC, OGG, M4A/AAC up to 20MB. Longer songs in lossless formats can exceed the size cap — converting to MP3 first usually solves it.
No. Your upload is used for the one separation you requested and deleted immediately after processing. Nothing is stored, shared, or used for training.