Louder without the crackle
Raising the volume multiplies every sample by the same amount. Measured in decibels, +6 dB doubles the signal's power and +10 dB is roughly what people hear as twice as loud. The quiet parts come up, and so do the loud parts — which is where a naive boost goes wrong.
Digital audio has a hard ceiling. A peak pushed past it is cut flat, and a flattened peak is the crackle you hear from a phone speaker played too loud. Every boost here runs through a limiter that catches those peaks and holds them just under the ceiling, so the body of the recording gets louder while the transients are tamed rather than clipped.
- Boost versus normalise
- A boost adds a fixed amount. Normalising measures the whole recording and sets it to a standard loudness. Boost when one file is simply too quiet; normalise when several files need to match.
- Fades are measured on the result
- A five-second fade out ends exactly where the file does, and a fade can never be longer than half the recording, so a short clip is never faded from start to finish.
- It cannot add detail
- Hiss, room noise and distortion captured in the recording come up with everything else. A boost makes a quiet recording audible, not cleaner.