Quote Hugh Robjohns:
I'll move this thread to the mastering forum as you'll get a more comprehensive response there.
However, my own thoughts on the matter are that once a track mix is complete there is inherently no further requirement for headroom within the digital domain and so there is an obvious argument for allowing peak sample values to reach 0dBFS. And most commercially mastered material does follow this idea.
However, this approach ignores the well proven fact that the analogue waveform reconstructed between the digital samples can, and often does, exceed the amplitude represented by those digital samples.
Given that pretty much every D-A converter these days employs some form of digital oversampling filter technology, these inter-sample peaks, as they are known cause obvious problems.
You can have perfectly legitiimate 0dBFS samples at 44.1kHz, say, but as soon as you try to oversample that signal in a digital filter the newly calculated intermediate samples may well need to describe amplitude values grater than 0dBFS... which is obviously impossible. The result is overload distortion in the digital domain and it is not that uncommon to find in commercial CDs.
Also, because of the similar digital filtering that is part and parcel of MP3 coding, similar overload distortion problems crop up regularly.
The best way around this is to use high quality oversampled digital metering to assess the true reconstructed amplitude of the signal, but if that is not available, then setting your maximum peak sample value to something between -0.25 and -1dBFS is a good idea and helpful safeguard. And no one is really going to get upset that your material is fractionally quieter than someone elses...
If they do, show them where the VOLUME CONTROL is on the ipod / amplifier / mixer!
Yet another post that will be framed and hung in the loo!...can I show this to my students?...."Dave, why do your mixes sound quieter than Muse?"...AAAAARRGGGHH!
My head hurts!