Fix clipped, distorted audio: De-clip or peak reduction?
Identify clipped peaks, evaluate estimated repair and leave output headroom.
A loud syllable crackles and the waveform top is flat. Turning down the volume leaves the distortion intact. Lowering gain reduces an already damaged signal without rebuilding its clipped shape.
Look for evidence of clipping first
Zoom into the distorted syllable. Flat tops together with crackle may indicate clipping. Other distortion exists too, and after earlier attenuation the clipped plateau may sit below 0 dBFS. Match the threshold to the actual defect.
De-clip estimates from the surroundings of a short defect
De-clip selects short sections using threshold and width limits, then estimates peak shape. Leave output headroom before Preview because repaired peaks can be taller. Recovery limit bounds the estimate; it does not measure the lost original.
Peak reduction when a spike is simply excessive
Peak reduction reduces excess amplitude above threshold rather than reconstructing anything. Use it to tame an intact but disproportionate peak. For uniform attenuation, Gain is more direct. Keep these tasks distinct.
Compare the same passage at comparable levels
A quieter result can initially seem less harsh. Check whether crackling actually improved or the whole word merely became softer. Review surrounding speech and new peaks, then listen through the full applied section after the short Preview.
Long overloads impose an information limit
Long flattened passages lose more information. Ease the repair or retain the original if it creates new clicks or odd timbre. A smooth waveform does not prove faithful restoration. A specialist sample can help assess irreplaceable material.
Try your own recording
Start with a short passage you can review carefully. Local editing and Preview are free; finished-clip export is paid. The editor’s price calculation shows the current cost.