I vaguely remember posting something like this some time back…
Normalize/Amplify across a multi-channel show such that the stereo mix-down does not overload.
This is coming up more and more with the Overdubbing documents. The only way to get a clean stereo mixdown is hunt and peck Group Amplify until you hit it. I know of no way to efficiently reduce the track gains in a controlled manner.
It’s a given that you can’t change the individual tracks because that affects the theatrical mix.
Koz
This can be seen as a major headache, or a minor inconvenience.
A major headache because it can ruin your show - a minor inconvenience because it shouldn’t ruin your show, in fact when approached the right way it will do no harm at all.
When mixing down, forget about numbers and meter readings - listen to what it sounds like. If it sounds good, it IS good. The only number that needs any consideration is “32”. More than at any other time, 32-bit float is a must. If the final level is a bit low or a bit high the Amplify effect will sort it out with no damage at all provided that all tracks are 32-bit float.
The minor inconvenience is evident as soon as one uses a multi-track audio program that has some sort of “master fader”. The addition of a master fader does make mixing down quicker, easier and more intuitive.
Please add your vote(s) if you’ve not done so already: Missing features - Audacity Support
One-step way to avoid clipping on multiple tracks: (25 votes)
Master Fader ability: (20 votes)
Separate mixing window (like Mixer Board but with a “Master-Fader” control): (5 votes)
Shortcut that moves all gain sliders by same amount (6 votes)
Master-Fader or Master Envelope track: (3 votes) Like the Envelope Tool but in its own track and applies to all tracks, used additional to and independently of Envelope Tool. On playback, fader position is interpolated (linearly in dB) between the track points and the result is added (in dB thus multiplied) with the current “envelope” value.
When mixing down, forget about numbers and meter readings - listen to what it sounds like. If it sounds good, it IS good.
Said the skilled musician. However most of us do depend on our instruments and tools and the vast number of people messing around mixing have no idea what they’re hearing.
I don’t see “Linked Normalize on Export” tool anywhere in there. The last one comes close, but I don’t like the word “Automatic” in the title. It suggests it’s constantly riding levels as I mix.
Master Fader is a terrific idea. That’s the two big red faders on the right of the console. Please note that the console metaphor extends to a small meter and overload flasher on each channel as well as the overload flasher on the two output channels. However I’m talking about two/three key clicks just before Export, not a phrase by phrase mixing tool. They’re separate.
Yes, I understand 32 is a requirement, but if LNoE did the math, scaled first and then exported, the user would get a slightly noisy show out of 16-bit that we could then tell him/her out to avoid instead of producing unrecoverable trash.
Koz
And all those years I was running Pro Tools in 16-bit 44.1k and producing trash!
– Bill
Tee-hee. Thanks for the fly in the ointment Bill 
ProTools is built on a different paradigm to Audacity in that operations on the audio are mostly done non-destructively in real time so the lack of support for float format audio in ProTools is a much less limiting. Also ProTools has a considerably more sophisticated mixing and metering section. As long as the original recording levels are OK clipping during processing usually does no harm to the actual audio data and the level can be adjusted lower to correct the clipping.
In Audacity when you change the audio it really does change the audio there and then, so multiple operations will be compounding rounding errors if integer formats are used in the tracks. If an operation to an audio track causes the peak level to exceed 0 dB, then unless float format is used the clipping will be permanent, whereas if float format is used the track can be “amplified down” to correct the clipping.
No I don’t like that idea either - it sounds like it will have all of the inherent problems of AGC (automatic gain control).
Assuming that everything is set to 32-bit float format, three steps before Export:
Yes I can see that something like “Tracks > Normalize Mix” would often be easier and quicker, but there would probably still be bear traps with non-contiguous audio clips.
I definitely like the idea of some sort of master fader (and have already voted for it).
Yes, I realize that. The downside (at least in Pro Tools 6LE) was the mixdown happened in real time.
No kidding! 
Yes and no … and we’re back to the discussion of whether or not Audacity does all internal processing at 32-bit float, which I contend it does.
No argument with that.
– Bill
Same in ProTools 7.
There’s a thread from somebody with a rock band who really needs to mix down many multiple tracks to stereo. He keeps getting really close through painful, detailed track adjustment, but so far no cigar.
What did we decide? Thread petering out isn’t going to do it. Koz
My vote is for a master fader.
This could work one of two ways.
It could work like a traditional mixing desk and scale the mix
channel audio → channel level sliders → master level slider → output
It could move all channel sliders by the same dB amount.
channel audio → channel level sliders → output
Method 1 has the advantage of familiarity, but has the disadvantage that metering could appear weird. Would a soloed track display on the meter with a pre master fader level or a post master fader level?
Method 2 has the advantage of simplicity and transparency - moving the master fader would visibly make all other faders move. The method could perhaps be scalable to allow sub-master controls that operate on groups of tracks.
My preference would be for method 2. Although it is less conventional for a mixer, it is probably less confusing for people that are unfamiliar with mixers. I would find the ability to “group” tracks particularly useful.
If Audacity were to have a master fader I would prefer that it operate per industry norms.
If soloed tracks are metered then they should be pre-track-fader and pre-master-fader as per industry norms.
As for the thread in question, all he has to do is Edit > Select > All then Effect > Amplify with a setting of -3 dB. Or he could reduce each track fader by 3 dB. The latter is picky and time-consuming but that is the nature of the beast.
– Bill
If a master gain control were added, I agree with Bill’s “industry norms”. The extra programming to add a master gain (and especially the GUI implications) are probably non-trivial. If, as I understand, using Amplify does not add artifacts (i. e. its math is all 32-bit float) then there is no need for said master gain. I rarely need to mix multi-track shows (but have done so quite a bit in the last month) and find Amplify sufficient unto my needs.
IMO Audacity would need a “proper” mixer board view to support a master fader, proper track soloing etc. We then move closer and closer to being Ardour.
– Bill
I’m not a C+ programmer but I suspect that option 2 (move all channel sliders by the same dB amount) would involve a lot less programming than option 1 (which I agree would probably need a “proper” mixer board view and would be a major project).
The code to “move all channel sliders by the same dB amount” is fairly simple. The designing and placing of a control in the GUI is a QA question first then a programming implementation chore – both of which are likely to be less than trivial.
How about the equivalent of the “double-click-on-the-gain-slider” dialog, but apply the relative adjustment to all tracks?
– Bill
or a “Link Gain Sliders” button on the Mixer Board?
As Edgar said, designing and placing of a control in the GUI is a QA question first. Do we want to start discussing it here? I would suggest that it may be better to discus it on a wiki Proposal page.
Coming back to the original topic “Normalize/Amplify across a multi-channel show such that the stereo mix-down does not overload”:
Both the Normalize and Amplify effects can calculate the required gain extremely quickly because the peak level of the selected audio track is already known to Audacity (the minimum and maximum values of each block file is listed in the AUP file).
In calculating the necessary gain to normalize a complete project, the peak levels of individual tracks is insufficient. The only way that the gain could be accurately calculated would be for Audacity to mix all of the tracks first, and this would make the process very much slower. This would be an automatic equivalent of:
Considering the slowness of this, how worthwhile is it?
BUMP
I have moved the relevant discussion from this thread to the “Pending Feature Requests” page of the Wiki.
But I agree with Steve that this cries out for a proper proposal - any takers from among you elves?
It seems to me that this would be an incredibly useful addition to Audacity’s armoury of tools for our mult-track users - and a good early-post-2.0 project to tackle.
(BTW I’m not volunteering for this one. 
)
Peter.