Buscar archivos
or drag and drop it anywhere in this area
No file size limit · minimum 10 seconds · 30+ seconds recommended for analysis
Files stay on your device — no uploads · no account · private processing in this browser tab.
Balanced — good default for most mixes
- stems rendered
- 2stems rendered
- signals combined
- 3signals combined
- files stored
- 0files stored
- lossless export
- WAVlossless export
What the separation measures
Azimuth discrimination
The stereo field is analysed per frequency bin to estimate how centred each component is. Centred content — usually the lead vocal — is attenuated in the instrumental and retained in the vocal render.
Spectral masking
A short-time Fourier transform separates harmonic and percussive structure, so bass and drums survive the centre reduction instead of being cancelled along with the voice.
Local rendering
Decoding, masking and WAV encoding all happen in this browser tab using the Web Audio API. There is no upload step, no queue and no account.
What it can and cannot establish
Separates cleanly
- Wide modern stereo mixes with a centred lead vocal
- Pop, rock and electronic masters with clear panning
- Tracks with dry or lightly reverberant vocals
- Studio recordings at 44.1 kHz or higher
Leaves audible bleed
- Mono recordings, where there is no stereo difference to use
- Heavy reverb, delay or wide doubled vocal takes
- Live bootlegs and heavily bus-compressed masters
- Low-bitrate sources with smeared high frequencies
Cómo funciona
Upload a track, our in-browser engine analyses the spectrum and renders two stems: the instrumental and the isolated vocal. Nothing is uploaded anywhere.
- 01
Upload
Pick a file or drop it in — decoding happens on your device.
- 02
Process
The spectral engine renders the result in this browser tab.
- 03
Download
Preview and save the output as a lossless WAV file.
Spectral engine
Frequency-domain processing tuned for voice and instrument content.
Fast
Rendered locally in this tab — no upload waiting or queue.
Free
Unlimited tracks, no account and no usage limits.
Analysis and separation results are estimates produced by signal processing, not definitive verdicts. Quality depends on the source recording.
How vocal removal works in your browser
A vocal remover splits a finished stereo mix into two useful parts: a lead vocal track and an instrumental (karaoke) track. This page does that entirely on your own machine using the Web Audio API — the file you choose is decoded in memory, processed, and offered back as a download. Nothing is uploaded to a server, so a full-length album track is as private as a voice memo.
Two families of technique are combined here. The first is stereo-field processing: most pop and rock mixes place the lead vocal in the centre of the stereo image, so subtracting one channel from the other cancels the centre and leaves the sides. It is fast and lossless-sounding on wide mixes, but it also removes bass and kick, and it fails on mono or heavily bus-compressed masters. The second is spectral separation, which analyses the time–frequency content of the mix and attenuates the bands where a voice dominates, keeping instruments that live above and below the vocal range.
Real separation quality depends on the source material. A modern, wide, well-mastered mix separates cleanly; a mono recording, a live bootleg, or a heavily reverberant mix will leave audible bleed. If you need drums, bass and other instruments as individual files rather than a single instrumental, use the stem splitter instead.
How to use the vocal remover
- 01Drop an MP3, WAV, M4A, FLAC or OGG file onto the upload area — or pick one from your device.
- 02Wait for the waveform to render; decoding happens locally and is usually instant for a 5-minute song.
- 03Preview the vocal and instrumental tracks and balance them until the separation sounds right.
- 04Download the instrumental for karaoke and backing tracks, or the acapella for remixing and sampling.
Frequently asked questions
Yes. All processing runs in your browser using your own CPU, so there is no per-file cost, no account, no watermark and no queue. The only practical limit is the memory your device can give the browser tab, which usually means files up to roughly an hour long.
No. Audio is read with the File API, decoded into an AudioBuffer, and processed in the page. It never leaves your device and is discarded when you close the tab.
Reverb, delay and doubled vocal takes are spread across the stereo image rather than being purely centred, so centre-cancellation cannot remove them. Highly compressed masters also blur the separation. Try the stem splitter, which models sources rather than the stereo field.
WAV for lossless editing and MP3 for sharing. Bitrate and sample rate can be adjusted in the audio converter if you need a smaller file.