Q&A: When Do I Normalize?
If you read yesterday’s post, you know that AASG has a new writer, Andrew Levine. He’ll be popping in from time to time with articles regarding recording and otherwise awesome audio stuff.
One of the questions we posed on the AASG Google Wave group was this: Would you, at any time in your production workflow, normalize your audio? The response to the question was good and definitely varied. Andrew decided to weigh in on the subject via this post, so read it and feel free to comment. If you comment with your email, I may just send you a Wave invite!
Regarding the question of when to normalize audio I’d like to put that in perspective with regards to my work-flow.
When I do the final down-mix of a project (mostly in the domain of classical music, jazz or avantgarde), with material at the final sample rate and a dither plug-in in the last slot of the master bus (to reduce to 16 bit for CD-A), I check the peak levels of the loudest passages to make sure they stay below -0.2 dBFS.
You need some headroom when you plan to possibly trans-code the audio, e.g. to an mp3, as peaks can crop up and you definitely don’t want to go above 0 dBFS.
If you plan on sending the file to someone for mastering, or if you are bouncing a file that is to be placed in a compilation of material from other projects, it is useful to preserve headroom in the range of 3 to 6 dB–without clipping the material! Bounce to 24 bit fixed or 32 bit float and don’t maximize levels before you know where this piece of music will fit in.
If I see that there are occasional (and few!) peaks that would either create overs or force me to lower the overall level there are two choices: manually and musically lower the level of the clipping passages (I’ll pick up the topic of “automated gain riding” in a followup) or insert a brick-wall limiter in the processing chain (before the dithering plug-in).
So, when do I personally normalize?
If I have a file that was rendered with some headroom and want to hand it on I normalize it (to -0.2 dBFS) before reducing the bit depth and transcoding it. If I want to send off a small group of similar files I’ll do the same, normalizing all audio to the same level, so that the recipient has an easier job. And that’s about it :-)
Loading ...
