Mastering for a specific format - ie: mastering for mp3 or mastering for iTunes - is an idea akin to the 70's era of recording, mixing, and mastering.
That idea says one did whatever one needed to during
those procedures, however odd or counter-intuitive they may seem, so
that the resulting vinyl pressing matches the artists' vision.
Apple's ‘mastered for iTunes’ philosophy basically follows what one should be doing when converting to mp3, or any lossy format.
the typically loud, pop genre master many clients still request today
(in 2012), when outputting to an mp3 delivery format, the output limiter
should having a lower ceiling set at - for example -2.0 dB, as opposed
to the usual setting for CD, .wav, or .aiff files of -0.1 or 0.0 dB (or
-0.3 dB which is what I generally use. Yup, I guess my masters are slightly
The reason for setting the limiter that way is that when rendering to a lossy format the resulting file generally will have peaks going past the limiter's ceiling - in this case -2.0 dB.
Let’s say you render to a 192k mp3 with the limiter settings I described above. You will find that when analyzing the file for the highest peaks you'll often see values of -0.9 dB, or -0.25 dB, etc. obviously 'dis-obeying' the limiter absolute of -2.0 dB.
If I don’t set the limiter ceiling at -2.0 dB the resulting lossy format files will have clipping distortion, hitting 0 dB.
Here it is again, in case you missed this:
Of course, setting the output ceiling of the limiter at -2.0 dB, requires some extra work in re-adjusting the levels to the input threshold. That’s so the final RMS levels of the mp3 or iTunes format, match that of the CD/wav/aiff version. If this isn’t followed, we’ll have a file that's 2 db quieter than the master CD file.
course, if we're starting out with, say, a 96 kHz 24 bit file, we'll
need to choose an excellent resampler and dithering plugin to
maintain that quality as much as possible. I have excellent results with
WaveLab’s Crystal Resampler. Voxengo’s r8brain PRO is another highly regarded resampling software.
Another procedure I follow when mastering for mp3, is to low pass filter the file at around 15 kHz (with a high quality filter of course) since many lossy formats don't have content above that.
I might do some extra mid-side EQ and 'soften' the side levels by 1 dB, or so, at that frequency range, if the stereo image is rather pronounced in that region to minimize any artifacts that might result.
This is the reason I charge an extra fee
for providing additional mp3 versions for an artists project should
they request it. A little extra time is involved to make them sound the
best they can be.
Naturally, the procedures I’ve outlined for mastering for mp3 won't create a precise match with the master file.
This ensures a better mp3 listening experience for all my clients' listeners.
Proud to be affiliated with these organizations: