Of all recording mistakes, clipping is the most frustrating because it almost always cannot be fixed afterwards. A perfect vocal, a one-off interview or a great take can be ruined by a few moments that are too loud. The good news is that avoiding it is simple, as long as you understand what headroom is and how to set gain.
What clipping is
In an integer digital system there is a maximum value, 0 dBFS. If the incoming signal would require a higher value, the converter cannot represent it and records the maximum available. The top of the waveform is thus "clipped" and flattened.
A clipped wave contains harmonics the original sound did not have. You hear it as:
- a snap or crack on peaks, if the clipping is brief;
- a rough, continuous distortion, if the signal stays above the limit for long;
- in severe cases, a smeared, fatiguing sound.
Analog and digital clipping
Analog circuits saturate too, but usually gradually: the waveform gets rounded. Digital clipping, by contrast, is abrupt and gives no warning. That is why a slight overload in an analog preamp can be tolerable, while at the converter input it never is.
Why it can't be repaired
Repair tools can plausibly reconstruct short clipped sections by estimating the shape the wave would have had. They work well on a few isolated peaks, but the result is still an interpolation. On long or heavily saturated passages there is nothing to recover: the information was simply never recorded.
Lowering the volume of a clipped file does not remove the distortion: it just makes it quieter.
Headroom: the safety margin
Headroom is the distance between the signal's typical level and the maximum possible. With 24-bit recording the available space is large, so you can leave plenty of margin without noticeably increasing noise.
Reasonable starting values:
| Source | Indicative peaks |
|---|---|
| Spoken voice, podcast | between -12 and -6 dBFS |
| Singing voice | between -12 and -8 dBFS |
| Percussive instruments | between -10 and -6 dBFS, with extra margin |
| Unpredictable live events | -12 dBFS or lower |
These values leave at least 6 dB in reserve for a sudden laugh, a harder hit or an emphatic expression.
Setting gain step by step
- Turn the preamp gain all the way down and place the microphone as it will be for the real recording.
- Ask the person to speak or play at the loudest level they expect to use, not their warm-up volume. Many people speak softly during the test and much louder once they focus.
- Raise the gain until peaks land in the indicated range.
- Take off another 3-6 dB if the source is unpredictable.
- Make a short test recording and listen on headphones, checking the waveform.
Where to look
Every interface has a peak indicator, often an LED that turns red. That matters more than the DAW meter, because it concerns the converter input. If the LED lights up, clipping happened before the signal reached the software, and lowering the fader in the DAW won't help.
Common mistakes
- Adjusting level with the track fader: the fader acts after recording and does not change what the converter receives. Gain is set on the preamp.
- Chasing a "maxed out" signal: that was sensible advice with 16 bits, it no longer is.
- Forgetting automatic gain: some apps and operating systems apply automatic gain control that alters levels. For serious recordings it should be turned off.
- Clipping in plug-ins: inside a floating-point DAW a peak above zero does not break, but a plug-in emulating hardware can saturate. Check the input levels of every processor.
- Output limiters hiding the problem: a limiter on the master bus keeps the export from exceeding zero, but it does not repair an input that already clipped.
A useful habit
In my work, at the end of a session, I quickly scroll through the waveforms of all tracks full-screen looking for flat-topped peaks. It takes a few seconds and lets you catch the problem while you can still redo a part.
For interviews, where repeating is hard, a practical solution is to record a second track of the same microphone in parallel with the gain 10-12 dB lower, if the interface allows it. If the first one clips, you use the second.
In short
- Digital clipping flattens the waveform and adds distortion that is almost always irreparable.
- Headroom is the space between peaks and 0 dBFS: at 24-bit you can leave plenty.
- Gain is set on the preamp, testing at the loudest expected level.
- The interface's peak indicator is more reliable than the DAW fader.
- For one-off events, a safety track at reduced gain is a simple protection.
