Signal Processing
What Is Clipping in Audio?
Quick answer
The water-in-a-glass analogy
Imagine filling a glass to the brim and then adding more. The water overflows — you can't fit more in the glass than it holds. Clipping is the audio equivalent. A digital audio system has a maximum value it can represent (0 dBFS — zero decibels relative to full scale). Any signal that would exceed that ceiling gets cut to the ceiling instead.
A normal audio waveform has smooth, rounded peaks. A clipped waveform has flat tops — the rounded peak was "cut off" at the ceiling and replaced with a flat line. That shape change is what causes the harsh sound.
Normal waveform
Rounded, smooth peaks
Clipped waveform
Flat tops — peaks cut off
Why clipping sounds harsh
The flattened waveform contains sharp edges — the corners where the rounded peak transitions abruptly to the flat ceiling. Sharp edges in a waveform correspond to high-frequency harmonic content, specifically odd harmonics (3rd, 5th, 7th...) that weren't present in the original signal.
These new harmonics are inharmonic relative to the musical content — they don't belong to the note being played. That's why clipping sounds harsh, buzzy, and unpleasant rather than just louder. The distortion isn't adding a musical quality; it's adding noise.
Digital vs analog clipping
Digital clipping is immediate and harsh. The moment a signal exceeds 0 dBFS, the waveform is cut to a flat line. There's no gradient — above the ceiling, the waveform is destroyed. The audible result is an unpleasant crack or buzz.
Analog clipping — from an overdriven tube amplifier or tape recorder — is softer. Analog circuits saturate gradually, rounding off the peaks rather than abruptly cutting them. The harmonic content produced is different: more even harmonics (2nd, 4th), which integrate more musically. This is why "tape saturation" and "tube warmth" are desirable in music production — they're forms of controlled, soft clipping.
The distinction matters: analog-style saturation can be used intentionally and sounds good. Digital clipping almost always sounds bad and should be avoided.
Intentional vs accidental clipping
Some producers intentionally clip audio for creative effect — particularly in electronic music and lo-fi genres, or when emulating the sound of overdriven equipment. Soft-clipping plugins and tape saturation plugins replicate the analog clipping character without the harshness of hard digital clipping.
Accidental clipping — an input signal recorded too hot, a master that exceeded 0 dBFS before export, a voice that peaked during an interview — is always a problem. The goal in recording is to leave headroom: record at levels below the ceiling (a target of -18 to -12 dBFS during recording is typical) so transients have room to breathe without hitting 0 dBFS.
How to avoid clipping
- Record with headroom:Aim for peaks around -18 to -12 dBFS during recording. You can always raise the level later; you can't unclip a signal that was clipped at recording.
- Use a limiter before export:A brickwall limiter set at -1 dBTP prevents any samples from exceeding that level. Streaming platforms also specify true peak limits (-1 dBTP is standard).
- Monitor your meters:Most DAWs show a red indicator when clipping has occurred. Check after recording; don't just assume the levels were fine.
- Watch for inter-sample peaks:When digital audio is converted to analog for playback, inter-sample peaks can exceed 0 dBFS even if no individual sample clipped. A true peak limiter catches these.
What to do if you have clipped audio
Once audio is clipped digitally, the information is genuinely gone. The flattened peak can't be restored because the original waveform data wasn't stored. Declipping tools exist (iZotope RX, Adobe Audition's diagnostics) and they can partially reconstruct the original peak shape using educated guessing — but results vary from "acceptable" to "still clearly damaged" depending on how severe the clipping is.
Mild clipping (a few samples occasionally touching 0 dBFS) is sometimes repaired well. Sustained clipping — an entire vocal that was recorded too hot — is difficult to salvage fully. Prevention is far easier than correction.
Last updated: March 28, 2026