Dithering with a plugin

[R C Jara]

Hello everyone, im mastering an album wich is going to be a CD. All the song mixes and the project are in 48k 32float.
Each song is in a different chanel with his own chain of mastering and everything goes to the Mix chanel which only have a Loudnessmeter.
I noticed that when i set the limiter ceiling to -1db and then i look the loudnessmeter in the same chanel it shows me -1db, but if i look the loudnesmeter in the MIX chanel it shows me -4db, the same happens if i export that song and look the wave: it shows me -4db of peak and its quieter than the same song that im mastering. To avoid this i just add 3db in the IN of the MIX chanel which solve the problem. Its ok to do this or there is an option to configure Adobe so i can avoid this step?

As i said this is gonna be a CD (44k,16b, WAV) so im going to use Dithering and i want to use the one included in the limiter of each song (Pro L2). My understand is that this process should be the very last step but, again, after the Limiter of each chanel i have to add those 3db in the IN of the MIX chanel to get the same volume.
If i do this is there going to be any problem?, Should i place the limiter in the Mix chanel instead?, Should i use the Adobe Dithering instead?

[SteveG_AudioMasters_ · Accepted solution]

The Adobe dithering is fine, and it really is the absolute last thing you want to do - adding 3dB will screw it completely, and will actually raise the noise floor (where the dither correctly would be) by the amount of gain you add. The most sensible way to do this is to do all the processing (including the sample rate change) first, keeping everything as 32-bit Float, and doing the dither as part of the save as a 16-bit file. As far as getting the level absolutely correct, the thing to do is to normalize each file to about -0.1dB before you do the 16-bit save, although you might need to alter that value slightly if you want some tracks quieter than others.

[R C Jara]

Thanks for your answer

 My question now is when doing  the Sample rate change, in Advance options i have quality (i put it in 100%, to my understand this gives me best conversion) and Pre/Post Filter. This filter is supposed to prevent aliasing  but isnt  that the propouse of the dithering? if i keep it ON wouldnt that mean that im doing a "double dithering"  kinda process?

[SteveG_AudioMasters_]

The quality factor actually doesn't make any difference - leave it at the default. It just comes up with a longer answer to an interpolation sum (more decimal places), but the used result is the same anyway! It doesn't give you a better conversion though - just one that takes longer. And it doesn't prevent aliasing - that's prevented anyway. Nor is any of this the point of dithering...

Dithering is about removing the 'certainty' of what happens at the least significant bit (LSB) threshold of the signal. If you don't do it, your signal falls off a cliff at -96dB and on a fade-out or reverb decay, that can actually be audible if you have the volume cranked up a long way.. Dithering fixes this by adding shaped noise to your signal at the LSB point. Also, by a strange coincidence, it can statistically let you have signals below the -96dB threshold - like reverb tails remain audible. It's only an issue with 16-bit and lower bit depths; 32-bit signals have enough coding depth to cope with anything that happens down around the threshold of hearing, so their inherent noise is coded accurately anyway - it's only when you reduce the bit depth that the 'cliff' effect becomes audible, and you have to take steps to prevent it happening. But it is important that you should only do this once.