Problem with Group Normalise- any ideas?

Question

Hi everyone- a bit of weirdness which might be my fault... or not. I've been playing with the Group Normalise function and it might solve my problem (creating a consistent volume with backing tracks I've recorded over many years and systems). I tried it once- and got encouraging results. Then I started tweaking the parameters and suddenly it won't process the multiple files I have selected... it just processes the one (seems the be the last alphabetically in the list). After restarting Audition a few times and trying to avoid or recreate the problem I actually had it NOT processing any of the selected files! The catalyst seemed to be tweaking the maximum Db level- once I tried that, things went wrong. But now I can't even "start all over again" to try and recreate the issue, because it's not processing anything.
This morning I tried again, hoping that a complete system restart might have dumped the bug or at least give me that one chance of getting things done, but once again the Group Normalise feature will only actually process the last/bottom file listed in the selection pane. I tried emptying my Audition TEMP folders, but there was nothing there anyway.
The feature "sees" all the selected files and analyses them, adding statistics to the list. It's just the final step that is failing. When I click on "Run Normalise" it applies processing to just that last, listed file.
I've been working on this too long now- so my memory's getting flaky, but I swear the very first time I tried this function it worked properly. But no luck since. Obviously, the 3.01 update could be applied, but there's no mention of this issue being fixed in the release notes.
Any ideas, anyone?
Cheers, Graeme.
Audition 3.0, SP3 Win XP. E6600 CPU 3Gb Ram. ASIO4ALL driver with on-board sound(temporary driver until I get a new firewire card with a TI chip for my MAudio FW410)
Cheers!
Graeme Hague
Author/Composer
Freelance Audio Engineer

Answer 1

Hi Graeme,
you wrote that:
> I'll investigate MP3Gain since everyone seems to think it's a good move.
No, fix it before going to mp3.
> However, I solved the problem in a straightforward way- back to Steve's "use
> your ears" concept.
Hmmm ... that is a luddite concept ....
> With 70 tracks to balance I chose the best of them all as a benchmark,
> then loaded them ten at a time into the Multitrack Mixer View and basically
> "mixed" them to equal perceived volumes with the benchmark track as a control.
As often previously mentioned the average with a 300 milliseconds window and no frequency compensation of any kind works best for me, only really weird stuff needs a manual lowering of level. What you need to do is to define what average level your canned noise should be at, AVG 22.5 works well for vinyl, AVG 30 to 25 for classical music, higher for larger ensemble with more bass and AVG 25 to AVG 17 is probable for church organ - depending on playing style, it can be extremely dynamic, ie. at AVG 35 - and pop. You do NOT need to go to full scale, what you need is predictability. In house notation used, I find it to be a good strategy to define what AVG rating to aim for with a project, also btw. for for multitrack mixing because compressor setting presets with input AVG rating defined makes a lot of sense to my way of thinking. And yes ... you do NOT have to tweak the track to fix, knowing the AVG rating tells you what input gain to apply in the compressor .... that is, if you DO insist on keeping the number of DSP actions on your audio low. Some of the time I prefer conceptual simplicity .....
> Whatever settings the mixer required I then applied in Edit View, zeroed everything
> to re-check and- voila! As close as they'll ever get
A visual evaluation can be a good tool, move the guide lines from -1 dB to -9 dB and they will become helpful.
A wise and well thought out design concept is that average values show up quite differently in the group normalize windowm, this entertains the user and poses new challenges so that working with a project doen't get boring. The function works fine IF you aim low, do not aim for 0 procent clipping, aim for no mentioning of clipping in any file, and skip the frequency compensation part of it, it uses a fletcher-munson compensation without telling you what playback level to aim for. It could have been useful if expected playback sound level was enterable as parameter ....however, skip that part of it and it will work. I have encountered file combinations that lead to a numeric overflow when using A2, dunno how it is in A3, I need to make my mind up about what box to install it on first ....
   Kind regards
   Peter Larsen.