Isolated Vocal
The singing lifted out of the mix, useful for remixes, practice and transcription.
◆ VOCAL REMOVER
Split a song into two tracks: the singing on its own, and the instrumental behind it. Use the instrumental as a karaoke backing track, or keep the vocal for a remix. The separation runs on Bubixo's own server, not an outside AI service.
After separation
track-vocals.mp3 · singing only
track-instrumental.mp3 · karaoke
What the tool does
The singing lifted out of the mix, useful for remixes, practice and transcription.
The backing track without the voice — the karaoke version people usually come here for.
MP3 at 192 kbps keeps the files small; WAV is lossless when the stems are going into an editor. Take both stems in one ZIP.
Use cases
Sing over the instrumental of a track that has no official karaoke version.
Mute the singer to rehearse vocals, or keep only the vocal to learn phrasing and timing.
Start a remix or a mashup from a clean acapella instead of a filtered mix.
Use an instrumental bed under narration without the original vocal fighting your voice.
How it works
The classic way to remove a voice was to subtract the left channel from the right, cancelling anything mixed to the centre. It removed the vocal, but it also removed the bass and most of the drums, and it failed completely on anything not mixed dead centre. The result always sounded hollow.
This tool uses an open-source source-separation model that was trained on music. It listens to the mixed track and reconstructs each part as its own recording, so the bass and drums stay in the instrumental where they belong. The model runs on our own server — your track is never sent to OpenAI, Google, AWS or any other outside service.
Privacy and limits
⚠ Good to know
Separation is heavy work. Expect roughly three quarters of the length of the track, so a four-minute song takes about three minutes. Only one job runs at a time, so if someone is ahead of you the page shows your position in the queue and starts automatically when your turn comes.
Quality varies with the recording. Modern studio productions separate very cleanly. Dense mixes, heavy reverb and delay, live recordings and arrangements where the voice shares a frequency range with the instruments are harder, and you may hear a faint trace of the singer in the instrumental. Nothing can be perfect, because the original separate recordings no longer exist inside a mixed file.
Frequently asked questions
Yes, with no account. There is a fair-use limit of five separations per day per person, because each job takes several minutes of server time.
MP3, WAV, M4A, FLAC, AAC and OGG, up to 50 MB and 10 minutes.
The tool returns vocal and instrumental, which is what almost everyone needs. The underlying model can produce four stems, so separate drums and bass may be added later.
Only one separation runs at a time so the server stays responsive for every other Bubixo tool. Your position is shown on the page and the job starts by itself.
For most modern recordings, yes. Listen to the instrumental before you rely on it for a performance — you can play both stems in the browser before downloading.
Start now
No account, no watermark and no payment. Upload a song and get both stems.
Bubixo audio tools are free to use. Files are deleted automatically after 24 hours.
This page belongs to Voice and Speech Tools, where every tool in the category is listed. These are the ones people usually open next.