Audiobus: Use your music apps together.
What is Audiobus? — Audiobus is an award-winning music app for iPhone and iPad which lets you use your other music apps together. Chain effects on your favourite synth, run the output of apps or Audio Units into an app like GarageBand or Loopy, or select a different audio interface output for each app. Route MIDI between apps — drive a synth from a MIDI sequencer, or add an arpeggiator to your MIDI keyboard — or sync with your external MIDI gear. And control your entire setup from a MIDI controller.
Download on the App StoreAudiobus is the app that makes the rest of your setup better.
lowering gain staging across track without losing relative levels/ preserving volume relationships
Hello!
I have been working on perfecting a mix all week and now have it sounding pretty good, however I have some tracks that are peaking, this isn't an issue because I have just lowered the master so everything sounds good pumping out
BUT
the clipping tracks pose a problem- when I freeze the track, the clipping tracks are radically reduced in volume
I would like to lower the volume of all of the tracks so that they remain in their current balance so that when I freeze tracks the mix does not become out-of-whatk
- but if my understanding is correct DB do not work linearly like this....?
If I lower every single track by -5 DB - will that preserve the volume relationships between tracks OR do I need to do some sort of calculation to determine relative volume changes along a logarithmic scale?
Also- if I switch the cubasis project to 24 bit audio- this would not improve the situation correct? Is my understanding correct that in order to completely avoid the possibility of clipping within a digital audio file that you need to use 32 bit floating such that you are working with an 'infinite' volume scale?
Comments
The clipping tracks should only get reduced in volume when freezing if you have normalization turned on.
Do you have any clue by how much the tracks are clipping?
If the 'clipping'(distortion/overdrive etc) is part of the 'sound' you want to preserve it's better to use a plug-in to 'clip' the sound while keeping the output below clipping level.
Cubasis always uses 32-bit floats for it's internal calculations so switching to 24-bit would not help you in this case...
@Samu Thank you for this!
I do not have ‘copy mixdown settings for freeze’ enabled so I am assuming I do not have freeze normalization enabled either as I do not see a place to set that other than within the mixdown settings?
I am clipping by 18db pre fader, by 7db post fader
What do you mean by 'clipping by 18db pre fader, by 7db post fader'
Freezing normally should not change the level of the track. Are you sure that it is?
If something is clipping +18db at any point something is f'ked up unless it's intentional gain boost at some stage and in those cases it should pass thru a limiter of some kind to keep the levels under control...
@samu it has a bunch of saturation on it so yes it’s intentional but I have all the levels at the master turned down so by the time it gets there it isn’t actually clipping...because of this I was thinking it shouldn’t have a limiter on it
Do you have any thoughts about what I was saying about normalization in the freeze/not having ‘use mixdown settings for freeze’ activated?
Also- if I do go to equally reduce the volume of all of the tracks, any thoughts about what I was saying in my original question about if reducing every track by 5db would preserve the volume relationships or if it would throw their ratios off?
@espiegel123 yes it’s definitely freezing in this instance...running Cubasis 3.1.3.... in Cubasis you can put inserts as pre fader or post fader, I placed an Analyser before the fader volume adjustment and got one peak reading, and then moved it to after the fader volume adjustment and got a different peak reading- I felt the need to mention this in the event that Cubasis renders the freeze files after the signal passes through the fader or something counter-intuitive
Either I am misunderstanding what you are saying OR you may be misunderstanding about accomplishing saturation. If you mean that the DIGITAL signal is hitting +18 db -- then that is a problem because you are not just saturating with your saturation but also digitally clipping which is something else entirely. And no amount of lower signal after it is digitally clipped will un-clip it (it will be quieter but still clipped). Usually when you apply saturation (or any effect to a signal) you make sure that it doesn't digitally clip. You can apply tons of gain in a plug-in and (in the plugin) adjust the output volume so that you don't digitally clip the signal.
If you aren't sure what digital clipping is, a discussion is worth having about that.
@espiegel123 i thought I was verrrrry familiar with the concept of digital clipping but i thought that most plugs process in 32 bit floating so that the only time that would occur would be the ceiling of a track at the time of bouncing to 16 bit or going through some sort of analog modeling plug?
I’m not sure what you mean when you say DIGITAL signal, when I put the 4pockets peak analyser on the track pre fader it gives me the +18db peak after the signal is passed out of FAC bandit and some other plugs
In terms of simple gain changes, dB's are linear (they're just a simple ratio expressed as a logarithm), but if you have any nonlinear behavior (clipping/saturation/compression/limiting/etc.) you can mess up where your signal is compared to the onset of the nonlinear behavior - for instance if you reduce your signal level before it enters a compressor you would need to reduce the compressor's threshold as well and similarly with saturation/intentional clipping/etc.
Note that it's at least possible an application may freeze things at a lower, fixed bit depth and cause clipping even if it's processing at 32 bit float. Historically I know some DAW's have had options to do this.
@annahahn : while it is true that 32-bit floating point gives tons of headroom -- ultimately when you save to disk you lose that headroom -- in general it is worth keeping your reference at 0 db so you don't have to worry about that transition from forgiving internal processing to saving to disk. There isn't really any benefit to your using such "hot signals". I'd recommend generally adjusting your faders so that 0 db is your reference and you won't have to deal with figuring out strategies for compensating. What matters ultimately, is the relative volumes of your tracks.
If the audio is saved to disk at 16bits it would either have to be reduced in volume or clipped. It’s not a given that it’s saved at 16bits by any means though. It may be, but there’s nothing saying it has to be for a track freeze.
@annahahn - just lowering faders by a fixed amount will definitely affect the relative balance of channels. The best you could possibly do would be to adjust each of them by a fixed percentage of their current settings individually. That would be difficult and even then wouldn’t be free from influencing the mix. Sorry.
Lowering the faders will maintain the balance unless there is non-linear processing going on somewhere. Adjusting gain is literally just multiplying each sample by the same value.
In any case, getting into the habit of targeting 0 dB as one’s reference will be beneficial and tend to avoid likelihood of issues like this where an app may use intermediate files. The big advantage of 32-bit floating point is reducing digital clipping when cascading dsp effects.
There isn’t any benefit that I can think of to making +18 dB one’s reference.
Fully agree. There is no benefit to going above 0dBFS unless you are deliberately trying to clip. It's just going to make nonlinear plugins clip and also clip when you leave floating point for any reason. And in floating point the signal to noise ratio does not improve with higher levels anyway since the noise floor is relative to the signal, not 0dBFS.
The original question was whether lowering all the faders by 5db would change the mix. It will. A fader at 0db lowered by 5db will reduce the volume by a much smaller percentage than one that is much lower. Say our scale is 0 to -100. Lowering a fader at 0 by 5db is a 5% decrease. Lowering a fader at -90 by 5 is a 50% decrease. Lowering a fader at -95 is a 100% decrease.
Please don't take me to task over the actual numbers I used. I know db aren't linear, etc. etc. it's just an example.
I won't argue the point if you still don't agree.
The point is this is incorrect. -5dB equals a ratio of about .56x*. So you're just multiplying every sample in each track by .56 and if you look at it that way you should see that the relative levels are the same.
It's tricky because dB's are logarithmic and that changes the math. For instance, adding dB's is actually the same as multiplying the corresponding ratios - that's one of the reasons they're convenient to use. I think it's actually easier to understand what's happening terms of ratios, but in practice dB's are very convenient to use.
*For anyone interested, these are the conversion formulas for the type of dB's used in audio:
dB's = 20 * log10(ratio)
ratio = 10 ^(dB/20)
I didn’t know that. Thank you for the education. You explained it well and I learned something.
Still I find it hard to accept that just lowering all the faders by 5dB would result in a mix that sounds the exactly same but quieter. Could be the case, but intuitively I just can’t accept it. Plugins, particularly compression and distortion react differently with different input levels. If they’re all pre fader, then Ok, but they might not be.
But it’s all academic anyway. Trying it is the best and only way to know.
For entirely linear stuff, reducing each track by 5dB and then mixing them is exactly the same as mixing them and then reducing by 5dB. It's literally just (.56 x "A")+(.56 x "B")+(.56 x "C") = .56 x (A+B+C) (i.e. the Distributive Property) if there's no other processing going on. If there's other processing involved, the question is if it conforms to the Commutative/Associative/Distributive Properties (aside from low level rounding errors).
Correct. I mentioned in an earlier post that any nonlinear processing like the types of plugins you mentioned will behave differently if you change their input levels.
Yeh, thanks @drewfx1 - great info. When I said I couldn't accept that it wouldn't affect the overall mix, I was referring to only the interaction with plugins that would change based on the input signal level, which is the case with virtually every track I make. I wasn't referring to the basic concept of lowering the db level without any such interaction. I get that now thanks to you. 👍🏼