Some info I found from dubstudio. Might help some peeps "What is the ideal amount of headroom to leave on a track before sending it to be mastered? Before answering that question perhaps we need to look at what headroom actually means. According to wikipedia "headroom is defined as the amount by which digital full scale (FS) exceeds the permitted maximum level (PML) in dB (decibels)". When it comes to mastering though, headroom is essentially a buffer zone to ensure our work can be carried out smoothly. When it comes to the headroom needed for mastering purposes, there are four main points to understand: 0dBFS, the peak level, the RMS level and the crest factor. 0dBFS is fairly simple, its the maximum possible digital level the signal can reach - a signal that would exceed 0dBFS simply overloads the system and leads to distortion. We use a peak level meter to measure how near a signal gets to 0dB, so we can avoid overloading the system. We use an RMS meter to give us some rough visual feedback on how loud the audio is. The crest factor is simply the difference between the peak and the RMS level. The image below shows a meter with peak level (thin vertical lines), RMS (medium vertical lines) and crest factor (thick vertical lines). From this we can tell that the RMS is about -15dB and the signal is peaking at around -3dB so the crest factor must be about 12dB. This particular meter plugin (called the TT Dynamic Range Meter) is available to download free from Brainworx (we can ignore the horizontal correlation meter for the purpose of this blog). TT Dynamic Range Meter, showing peak level, RMS level, and crest factor So far so good, if you are getting this sort of reading when you measure the levels of your music, we can work with it. Some producers prefer to mix to a much higher level though, so we often see a far higher peak level. In the image below, the track has now been limited so the RMS is around -9dB, the peak is hovering around -1dB (with a maximum of -0.2dB because we set the output ceiling of the limiter to -0.2dB) so we have a crest factor of about 9dB. The track is much louder and less dynamic - we can still work with it but we may need to turn it down before mastering it to avoid overloads at some point in our signal path. TT Dynamic Range Meter, showing peak level, RMS level, and crest factor Some forms of music require a much higher level than this, and to achieve the desired result we need to use heavy limiting. Some clients send material to us that has already been heavily limited, but we try to discourage this, because its really not necessary, and can severely restrict the work we can do. The image below shows a lot more red than the others! That's because the peak is not just hovering around -1dB, its crammed up against the -0.2 dB output ceiling. The RMS is only about -5dB, so the crest factor is only about 5dB. Its not overloading because the limiter is doing its job, but it has lost a lot of dynamic range. We would certainly have to gain this material down a lot to be able to work on it, so there is little point sending pre-masters that are this loud. TT Dynamic Range Meter, showing peak level, RMS level, and crest factor Finally I would like to show another image to illustrate why we need to be aware of all these four aspects (0dBFS, peak, RMS and crest factor). This image is the same as the last image, but this time the output ceiling on the limiter has been lowered to -6.2dB, so the whole track is 6dB lower in level. The peak level goes no higher than -6.2dB, and the RMS is about -12dB, but the crest factor is still the same (about 5dB). TT Dynamic Range Meter, showing peak level, RMS level, and crest factor This shows that what is really imporant is the crest factor, because although the level never exceeds -6.2dB, the audio has still lost much of its dynamic range. In theory there is plenty of headroom here, but in reality this audio is so heavily limited, that it makes little difference how much headroom there is. In summary, headroom is a useful term to describe the level of a track, but it must also be used in conjunction with peak and RMS levels to make any sense. "