Audio leveling backing tracks?

[Ian MacColl]

I'm interested to hear how folk automate pre-production audio leveling of backing tracks?

This is prompted by Paul's ffmpeg script on the Programmatically set trim (vs level) conversation but I thought it was worth starting a new conversation rather than hijacking that thread

Many shows I work on are based on backing tracks: music, choir and dance concerts, fashion shows, and so on

Dynamic range is often wide, sometimes inherently, say, for classical music, and sometimes because the track is a mash-up, say, for dance tracks

Normalising gives consistency between tracks but doesn't help within tracks so I'm interested in how folk manage wide dynamic range within backing tracks

Obviously it's possible to ride faders on a console, and QLab's integrated fade is also useful, but both of these are relatively hands-on in performance or preproduction respectively, and I'm interested in automation to get closer to set-and-forget

One possibility is applying compression in performance on a console or in QLab via the AUDynamicsProcessor audio effect

A related option is pre-processing tracks with compression as well as normalisation using, say, Audacity or ffmpeg or Logic Pro or GarageBand etc...

And a step further down this path is auphonic which is intended for podcasts and similar spoken content but has options for leveling music as well, removing the need to decide compressor settings such as threshold, ratio, attack, release, and make-up gain

I've gone pretty deeply down this rabbit hole, using r128x-cli referred to in the Level Playing Field QLab Cookbook recipe to contrast and compare different approaches

What do others do to level backing tracks? What methods or configurations work well?

[Paul]

First off, if a particular track I'd been sent had widely varying levels within it, I would check with the originator if this is an artistic choice and necessary for that particular piece. It may be that it was just a hurriedly put together mash up of music with different levels. 

If it's an artistic choice, then obviously leave it until any rehearsal. I  might prepare normalised or compressed versions in advance, if I thought that might help.  

If it was a problem at rehearsal / first performance and normalising the track didn't fix it, then putting a compressor / limiter on the mixing desk would be my preferred solution; doing it on the desk generally makes it easier to adjust. If nothing too severe was required in terms of limiting then it would be left for the whole show otherwise create a desk scene and recall the scene for that piece. But I really hate it when a severe limiter distorts the track, though accept you sometimes need to protect the PA from some DJs!

If there was no rehearsal / only one performance then I wouldn't worry about it, giving feedback to the director/originator if asked. Hopefully people learn from their mistakes and there is only so much you can do in a rushed production schedule.