Normalizing separate tracks before mastering

[Michaelm]

Hi all, I was wondering about your thoughts on normalizing certain tracks before mastering. What I mean is, though I have some of the equipment, I'm certainly not going to be mastering myself. I don't even want to learn that badly. Really I want to send all my mixes to a professional mastering studio who know what they're doing. Now I realize that most mastering engineers do NOT want us to tamper with, compress, EQ or limit the final stereo mix. But what about normalizing weak tracks during mixdown as a way of helping the mix? Would this make mastering any harder later?
Any input greatly appreciated!

 

[lowland]

Originally posted by Michaelm:
Hi all, I was wondering about your thoughts on normalizing certain tracks before mastering.

Hi Michaelm:

I'm sure others will have their own opinions on this but if you were one of my customers I'd say 'don't worry about it, there's no particular advantage': suppose you _had_ carefully normalised your material so it all peaked close to 0dBFS, there would still need to be gain changes during the mastering process, just as there would be if the tracks weren't normalised - time taken to adjust the gain in both cases would be very similar, so why do it twice with normalised material?

I think there must be a question mark over the extra DSP layer as well, and let's not forget that unless the artist is someone like Extreme Noise Terror it's fairly unusual for all tracks in a project to peak to the same level on a CD - generally, acoustic material sounds louder than electric and ballads are louder than rockers - both those types of material will likely need to be pulled back to make the whole album knit together in a pleasing way.

I advise clients to give me mix files untouched from when they were mixed whenever possible; by all means apply your own DSP of whatever flavour you fancy on copies to help get across a direction you'd like me to look in, even to give me an idea of the final loudness you'd like, but let's do the actual work on a neutral benchmark.

 

[Michaelm]

Thanks for the response. Just to be sure, when I say "track" what I mean is an indidual track in the song, not the entire song. I was thinking of track as in the vocal track, or the bass track, etc. So I've been thinking of normalizing some of the drum tracks in one of my songs, but not the vocals, guitars, bass or keys or flute. Would your answer be the same?

 

[lowland]

Originally posted by Michaelm:
Thanks for the response. Just to be sure, when I say "track" what I mean is an indidual track in the song, not the entire song

Offhand the only time you might want to do this is if an individual part had been recorded at such low level that channel/fader gains are insufficient to bring it in line with the rest of the mix, but that seems unlikely - if you're working 24bit you'd have to reduce gain by more than 20dB to get the eqivalent of a 16bit file, so you can usually adjust other part gains downward to compensate for quieter parts without fear of any major noise floor or dynamic range penalty.

I often ask customers not to bust a gasket to get 24bit mixes any closer than within shouting distance of 0dBFS: it's not necessary to have mixes peaking to, say, -3 or less as you might with a 16bit DAT machine, and the consequences of inadvertent clips could make that redundant anyway.

Otherwise the principle of my earlier post applies - why have additional DSP when it's not necessary? How does normalising make a 'weaker' part stronger? Final overall gain will be adjusted in the mastering process. If you don't have it already, I'd get the Bob Katz book 'Mastering Audio' where these and other related issues are very well addressed.

 

[Michaelm]

Again, thanks Nigel for the response. That makes sense and I won't normalize as the tracks are absolutely loud enough.

 

[lowland]

You're very welcome. All the best with your project.