Clean Up Inverted Vocals

How to clean up inverted vocals

1. Upload the inverted vocal. MP3, WAV, M4A or FLAC, up to 10 minutes. Sign in with Google the first time.
2. Press Clean up the vocals. The button shows the price before it runs: 1 credit for every minute. New accounts start with 5 credits. Most files take 1 to 3 minutes.
3. Listen, then download. You get the cleaned vocal back. Click a reel to hear it, and download each file as an MP3 at 320 kbps. If it fails, your credits are returned.

What is an inverted vocal?

Some acapellas are not made by an AI separator. They are made by phase cancellation. You take the full song and a version of it with no vocal, which is usually the official instrumental, flip the phase of the instrumental, and add the two together. Everything the two versions share cancels out and what is left is the voice.

It is a neat trick when both files line up exactly. In practice they rarely do. Different mastering, compression, tiny timing gaps or lossy MP3 versions mean the cancellation is not perfect. The vocal that comes out has leftovers in it: faint ghosts of drums or synths, a watery or swirling sound, pieces of the instrumental that did not cancel.

What this tool does

This model is trained for that specific problem. Give it an acapella from phase inversion and it tries to take out the leftover artifacts and leave the vocal.

Who uses it

Remixers and bootleggers who have an official instrumental and a full track and made an inverted acapella. Mashup artists who need a vocal that holds up when other music is placed beneath it. Producers sampling a voice that still has a faint beat underneath.

What works well, and what leaves traces

The cleaner the original cancellation was, the better the result. A vocal with a few light artifacts cleans up nicely. If the instrumental was badly aligned, the inverted file may contain large pieces of music that the tool cannot separate from the voice. Bits of the voice that were removed during inversion cannot be brought back.

If you did not make the vocal by inversion, the Acapella Extractor is the better starting point. For dull or thin vocals from low-bitrate sources, see the Audio Upscaler.