Vocal Remover & Karaoke Maker
BetaThis vocal remover (Beta) makes a karaoke instrumental from a stereo song, or keeps only the vocals as an a cappella, in a few seconds and without an AI model to download. It removes what is mixed in the centre of the stereo image within the vocal range, so bass, kick drum and the stereo width stay. Everything runs in your browser; the song is never uploaded.
- Runs on your device
- Web Audio
- Works offline
Updated
This tool is in beta: it works, but it is heavy for some devices and results may need a second look.
Drop your video or audio file here
How to remove vocals from a song
Step 1: Drop one stereo song (MP3, WAV, M4A, FLAC) or a music video on the box above.
Step 2: Under Keep, choose Instrumental for karaoke, Vocals for an a cappella, or Both to get two files.
Step 3: Pick MP3 or WAV under Save as, then press Separate.
Step 4: Play the results on the cards and download the stems you want.
Good to know
Formats
How it runs
- Web Audio
- Audio is processed with the browser’s built-in Web Audio engine, sample-accurate and fast.
Is this an AI vocal remover?
No. It uses centre-channel extraction, a signal-processing method: most pop, rock and hip-hop mixes put the lead vocal exactly in the middle, equally loud in both speakers, while instruments are spread left and right. Removing what both channels share in the vocal band takes out the singer. AI separators listen for the sound of a voice instead, which isolates more but needs a large model and a strong device.
| This vocal remover | AI stem splitters | |
|---|---|---|
| How it finds the vocal | What is identical in left and right, strongest from 180 Hz to 9 kHz | A neural network trained on voices |
| Start time | Instant, nothing to download | Often a model download of hundreds of MB, or an upload |
| Mono songs | Cannot separate | Can separate |
| Reverb and backing vocals | Usually stay in the instrumental | Mostly removed |
| Bass and kick drum | Kept (excluded from the vocal band) | Kept |
| Privacy | Song stays on your device | Depends on the service |
How does a vocal remover work?
The song is split into thousands of short, overlapping slices and each slice into frequency bands. In every band the remover compares the left and right channels: sound that is the same in both (the centre) is treated as vocal, sound that differs is treated as music.
A weighting curve limits this to the voice range: it is fully on from about 180 Hz to 9 kHz, fades out between 180 and 90 Hz (and is off below that) so the bass and kick survive, and eases off above 9 kHz to keep cymbals bright. The centre part becomes the vocals stem and the rest the instrumental. Both are rebuilt into normal stereo audio and saved as MP3 (192 kbps) or WAV.
Which songs work, and which do not
Results depend on the mix, not on the genre. Check these before you blame the tool:
| Song or file | Result |
|---|---|
| Modern studio pop, rock, hip-hop, EDM | Lead vocal clearly reduced; usually good enough for karaoke |
| Songs with lots of vocal reverb or delay | Dry voice removed, a ghost of the echo remains |
| Double-tracked or wide backing vocals | Those parts stay, because they are not in the centre |
| Centre-panned solo instruments (sax, synth lead) | Removed along with the voice |
| Mono recordings, including many releases from before the late 1960s | Refused: there is no stereo difference to work with |
| “Stereo” files with identical channels | Refused for the same reason |
| 5.1 surround audio | Mixed to stereo first, then separated |
Why can I still hear some vocals?
Because not all of the voice sits in the centre. Reverb and delay effects are usually spread across both speakers, backing singers are often panned left and right, and some producers widen the lead vocal itself. Those parts look like “music” to a centre-channel remover and stay in the instrumental. Compressed MP3s at low bitrates can also blur the left/right detail the method relies on, so start from the best-quality file you have.
Make a karaoke version or an a cappella
Karaoke version
Choose Instrumental. If the original key is too high or low for your voice, shift the instrumental with the pitch changer; the tempo stays the same.
A cappella (vocals only)
Choose Vocals. The result keeps the centre of the mix, which is mostly the lead voice plus anything else panned dead centre, such as a snare or bass note in the vocal range.
From a music video
Drop the video itself. The soundtrack is separated and you get audio files back; the picture is not included in the result.
Why use this karaoke maker
- Instant. No model download and no queue; a typical song is separated in seconds.
- Private. The song is decoded and separated in your browser, never uploaded to a server.
- Both stems at once. Get the instrumental and the vocals as two files from one run.
- Honest. Mono files are rejected with a clear message instead of returning a useless result.
Tips & troubleshooting
Start from a WAV, FLAC or high-bitrate file. The cleaner the stereo image, the cleaner the separation, and the less of the singer’s echo is left behind.
Leftover vocal echo in the instrumental? Lower the upper mids a little with the audio equalizer (2–4 kHz) to make it less noticeable.
Making a backing track for practice? Slow it down without changing the key with the audio speed changer.
Only need part of the song? Cut it first with the MP3 cutter; shorter files separate faster on phones.
Removing background noise from a voice recording is a different job; use the noise reducer for that.
Vocal Remover (Karaoke Maker) FAQ
How do I remove vocals from a song?
Drop a stereo song on this page, choose Instrumental under Keep and press Separate. The remover cancels the lead vocal that sits in the centre of the mix and keeps the instruments. Download the instrumental as MP3 or WAV.
Is this an AI vocal remover?
No. It uses centre-channel extraction, a signal-processing technique, so it starts instantly and needs no model download. AI separators can isolate more, including vocals in mono songs, but they need a large model or an upload.
Why can I still hear some vocals?
Reverb, delay, backing vocals and widened lead vocals are spread across the stereo field, so they are not in the centre and stay in the instrumental. Mono or near-mono recordings cannot be separated at all.
Can I remove vocals from a video?
Yes, drop the video file. Its soundtrack is separated and you get the instrumental and/or vocals as audio files. To put the instrumental back under the picture, use add audio to video afterwards.
Can I use the instrumental publicly?
Separated stems are still the original artist’s recording, so the same copyright applies as to the song. They are fine for practice and private karaoke; uploading or performing them publicly needs permission or a licence.
Can it work as an acapella extractor?
Yes. Choose Vocals under Keep to save only the centre of the mix, which is mostly the lead voice. Expect some centre-panned drums or bass notes in the vocal range to come along, and reverb to be missing, since that sits at the sides. For remixes, choose WAV to avoid an extra MP3 step.
Why is my song rejected as mono?
The remover needs differences between the left and right channels to tell the voice from the music. Mono files, and stereo files whose two channels are identical, have no such difference, so the tool stops with a message instead of producing a silent or unchanged track.