Headroom is not a property of sound – it is a property of the equipment that processes sound.
Headroom is a measurement of how loud the peaks of a sound can go above the 0dB reference point before the equipment starts to distort. In digital systems, the headroom is usually exactly 0dB (unless you adjust your meters and gain staging so that your nominal level is lower than 0dBfs). In analogue systems, the headroom usually depends on the quality of the components and the way they have been calibrated. Many analogue systems have a headroom of around 18dB, although this can vary considerably depending on the intended purpose of the equipment.
Here’s the key – audio equipment usually behaves according to the peak level of the sound, but we perceive based on the average level. Sounds with a lower crest factor can be pushed louder than sounds with a high crest factor. Therefore, a lot of the effort in increasing the potential loudness of a sound is focused toward using sounds with a low crest factor and reducing the crest factor of existing sounds.
If you thought that was complex enough, it’s only half the story.
It’s a similar situation with frequency, although its much easier to understand. Quite simply, audio equipment treats (or tries to treat) all frequencies as equal. But we don’t perceive all frequencies as equal. Generally, we are more sensitive to upper-mid frequencies (roughly 1kHz – 5kHz). For sounds of
equal recorded volume level, sounds that have a greater concentration of energy in the upper midrange will be perceived as louder than sounds with their energy focussed elsewhere.