16 May 2026 | 6 min read
How to fix clipped audio, and how to tell when it is unfixable
Clipping is missing data, not loud data. Some of it can be reconstructed convincingly, some of it can only be hidden, and knowing which saves hours.
A clipped sample is one the system could not represent because it went past the maximum value, so it was flattened to the maximum instead. The waveform grows a flat plateau where a curve should be, and that plateau generates harmonics your ear reads as harshness or crackle.
Spotting it
- Flat tops on the waveform at exactly full scale, visible when you zoom in.
- A gritty edge on loud consonants, brass or transients.
- Peak meters pinned at zero dBFS for more than a single sample.
- In video files, a distorted camera scratch track while the recorder track is clean.
What actually repairs it
Declipping tools estimate the missing curve from the surrounding samples and redraw it. This works well when clipping is brief and isolated, and poorly when a whole phrase is flattened. Always reduce the file level a few dB first so the reconstructed peaks have somewhere to go.
What only hides it
Turning the file down does not repair anything, it just makes the distortion quieter. A low pass filter softens the harshness and dulls the recording with it. These are last resorts for a take with no alternative.
The better fix: another source
On a multitrack recording, the boom may be clipped while a lav is clean, or the mix track may survive when an isolated track does not. Split the poly WAV, listen to every channel on its own, and check before you spend an hour on repair.
The player here lets you solo each channel of a poly file and see its peaks, so you can find the usable source in seconds without loading anything into a DAW.