Question on mixdown process - turning the levels down

[hirszu]

Hello guys, I'm currently gathering info and tips on mixdown. (Ableton is my DAW)

So far I have found out that the master channel should peak at -6db tops to leave headroom for mastering (correct me if this is wrong). To do that, I obviously have to adjust every channel's level.

My question is, could I instead use the volume control of an Ableton instrument/plug-in, and if not, why? For example, instead of turning down the level of the whole channel, I could turn down the volume of Simpler or Operator, or play with the output of the Compressor which is the last in the channel's effect chain.

To illustrate my question, let's say I have a snare channel (snare sample in a Simpler) which I want to turn down. My question is, should I do it with the mixer's volume control, or should I do it with the Simpler's volume control. Or maybe with the output of the Compressor, if the snare's compressed? All these options will have the effect of turning the volume down, but which one is correct/the best?

Here's an answer I got in a Facebook Ableton group:

[+] Spoiler

Great question. I would recommend adding a Utility device to the end of each chain during production. This allows you to turn a track up or down, or to add automation, while leaving the track's main fader free for adjustment in the mixdown stage. I don't think there's anything inherently wrong with the method you suggested, but perhaps some reading on gain structure would help you wrap your head around it.

I hope you guys can add something though.

Cheers!

If this should be posted in the global mixdown & mastering thread, then sorry

[FarawaySkies]

You can turn down the volume however you like. Just don't do it by turning down the master fader, or putting a plugin on the master (such as a limiter) and turning down the gain. -3db on the master is fine too, -6db is great, but dont lose sleept if your peaking at -2.5db, but you have a good mixdown.

More me, if I need something turned down, I turn down the channel fader. Doesn't really make sense to go through the extra steps required to turn things down by selecting instrument outputs / compressor gains. Speaking of compressor gains, make sure the volume of the track remains the same whether you have the compressor enabled or not. Adjust the makeup gain on the compressor to achieve this, then A/B the track by turning on/off the compressor to ensure the volume is the same. Then adjust the fader if you need it louder or quieter in the future (also if things are too quiet/loud I like to use the ableton utility plugin and adjust the gain).

[fragments]

I always start as close to the source of the sound to turn things down. Starting with the VST instrument itself is a great idea. DAW faders loose resolution/accuracy as you turn them down further, so starting at the source is advisable for a practical reason.

[SunkLo]

Yes, it's better to do what you're suggesting. At least to a degree. Some non-linear plugins will require a certain level hitting them to get the desired effects. Distortions, compressors, etc. So you don't want to go too quiet, but if you're turning things down to a sensible level that should only improve the results you get from your plugins. Most plugs worth their weight will have input and output gain controls anyway so you shouldn't run into any problems.

Getting it right from the source means you're not hitting any plugins too hot, which despite 32 bit audio engines, can still cause some undesirable effects. For instance I'm writing a compressor right now that has the ability to add a fair bit of gain if it's set up in an odd way. As a safeguard, I'm going to clip the output at +6dB in case someone sets the timing circuit up weirdly and feeds a loud burst of sound into it. If you had your source instrument pegging at +6 you'd be getting the shit clipped out of your audio, which is probably not the desired effect. Keep things below 0 though and everything's golden.

[nowaysj]

You can answer this for yourself if you start with proper gain staging.

Yes, good idea to have your track peaking at -6db at the END.

Probably a good idea to start with whatever the main element is at around -12db. -14db is not extreme depending on how layered up your track is.

I start with kick around -12db, and build the track around that. Invariably at some point, I've got to nudge the kick up a db or two, or the whole drum buss for that matter and there is headroom to do it.

If you stick to these practices, you should be able to answer this for yourself.

I like the ableton group's answer, in theory. I personally recommend trying to keep your track faders at 0db until the very end, doing the FINAL mix. This will require you to judge volume ALL ALONG the process, and signal chain. If I've got my kick peaking at -12db, and then start processing etc, I want that fucker still peaking at -12db. Many MANY plugins will give you 2-4db gain to their presets. I think this is very dishonest and is an attempt to deceive you. Fuck that. Unless I'm specifically using a plugin to increase my gain, I want signal peak in to equal signal peak out.

I control volume at the source. If some device in the chain is adding or subtracting gain, I add it back with that devices output gain.

Everything else will make sense relative to that.

I read that you are just starting out.

Good on you for figuring this mixing stuff out first, as it will allow you to more quickly express yourself. I did it backwards, and in my opinion, a totally fucking stupid course to take.

[forbidden]

i just group all my stuff, shift click across the groups in the channel strip and bring the volume down on everything that way. tinkering with individual volumes on compressors and stuff for each thing will drive you made. mind you want to be doing this at the last phase before you bounce out, and if you're on top of it you shouldn't really have to do it at all. learn to keep an eye on your master level throughout the whole process.

[hirszu]

Thank you for all the answers! I've got to read all that a couple of times because this topic messes with my head a bit lol

I read that you are just starting out.

That's right, I've started 2 months ago and so far I have only been composing. I feel that the track I'm currently working on may be my first one, so I've began reading about this whole mixdown thing to do it properly. Didn't read anything about mastering yet though.

Meanwhile I've also received more answers in the FB group and I think I should post a couple of them, if anyone's interested as they add to the subject/put the matter in a different way.

[+] Spoiler

Changing the volume via simpler/vst would mess with gain staging of your other effects. If you reduce the volume of your sample or sounds at the instrument, then all the compressors and distortions after would be receiving less signal, making the effect not be driven as hard and changing the sound you had before. Best to reduce volume at the channel fader, or throw a utility at the end of your effects and reduce volume that way.

^ is correct. You should NOT adjust you mix downs by using the volume of the synth. This will especially throw out gain-dependant effects/utilities such as compressors and post-synth sidechains. You want the final gain adjustment to always be your channel fader. Don't even use the final compressor gain to do the mixdown. The channel faders are there exactly for that, that is their function! They allow quick access to a track gain without having to touch the signal chain, so get into that habit. PS don't even worry if some of your track gains are clipping; so long as your master is still peaking at -6, there is no problem!

What you are referring to is gain staging.
You can adjust volume on any effects and/or processors and instruments but ONLY if they are causing the final out put to be causing digital distortion which really shouldn't happen now a days.
As mentioned above gain related processors will be effected when doing so.
In the 10 + years I have been producing I have never needed to mess with gain in my processors or instruments.
I made a video about this a while back and showed a couple of ways to make sure your mix is good.
I personally drop my master output to -6db and mix my faders by ear. That's what they are there for.
Also I NEVER have anything on my master output channel.

easiest way of doing all of this (and a way of avoiding 20 people arguing about gain structure) is a tip i was given by one of my lecturers a few years ago. a good habbit to get in is to add a utility to each channel (it should be last in the signal chain) then control the gain using that. at least that way you can keep your automation as is but still make overall adjustments

There's just one thing that I'm not certain about. To sum it up, people generally suggest adding an Utility plugin at the end of the signal chain, or just to play with the faders. Now as far as I understand these are two different things, because the Utility functions as an input gain control and the faders lower the volume. Gain does not equal volume, that I know (though I don't really understand it yet). I've got to read more on the whole gain staging topic.
Anyway, if I get you guys, when using a compressor or any other plugin which brings (or may bring) the gain level up, I should make sure that the volume remains the same? Just like in your kick peaking at -12db example @nowaysj.

Cheers

[SunkLo]

Nah fuck all those ableton cats, pull your volume down at the source before you add any effects. Then any plugins you throw on top will be calibrated for that level. If you've already got a long chain of effects set up, you'd want to insert a utility at the end so you don't mess up the level relationships, but if you've just added the first instrument or sample to a track, get it right from the source.

Notice when you load a sample into a drum rack, look at the volume on the sampler. -12dB. Prime example of trimming at the source.

Also, utility gain and the volume fader do the same thing. The reason people suggest using utilities or other gain trims instead of your fader is because the lower you pull your fader down the less resolution it gets. This is just because of the logarithmic scale of decibels. Also, if you get everything trimmed to read -12dB on the meters with all your faders up, and then use the faders to actually do the balancing, the position of your faders will represent the state of your mix. If you've got your bass cranked up, the fader will also be pushed up, some quiet background noise will have its fader down. It's more intuitive to work that way when everything's calibrated to the same level instead of having arbitrary levels hitting your faders.

When setting a compressor, your main goal is to eliminate any level change. Toggle the bypass a bunch of times while setting the output gain such that the before and after sound are the same subjective volume. Not peaking the same on a meter, the same volume to your ears. The compressed sound will peak lower because that's what compression does, so if you try to match the previous level on a meter it's gonna be cranked a lot higher. Volume match by ear and then you can assess whether the compression is actually better or just tricking you with loudness.

That trickiness is part of the reason PPM meters aren't a good candidate in DAWs. You're only being told peak value and not average value. So balancing things by their peaks will mean compressed sounds have a louder average level and very dynamics sounds will have a lower average level. I'd like to see some DAWs implement RMS metering on a track basis instead of having to use other metering plugins. Reaper has a VU/PPM hybrid meter on its master channel but not the others.

[hirszu]

Thank you, that cleared it up a lot. No more questions at this point I'll just get down to work. Big ups!

Edit.
I'll just add this article which is a great read. Hope this helps future readers
http://www.soundonsound.com/sos/sep13/a ... headed.htm

[nowaysj]

Just for sport Sunk, let me point out how I mix, as it seems a little different (I am suuuuch a bad mixer, so like take it with a grain of salt) - I get everything set at the source, which is the same as you are saying, except I mix at the source too. If I've got a big sound, it will be big at the source, and will be enbiggened along the way, but I'll bring it into the mix in such a way that the fader is at 0. So it sounds right in the mix with fader at 0. Likewise for a small sound, I'll keep it small, it'll come into the mix small, and the fader is at 0. This seems slightly different to what you say. I have some final mixes where the faders are completely flat - I've done all the volume mixing in the device or the chain. Or most likely, I'll do subtle rebalancing with channel faders. Usually kick and snare up a nudge, hats down a nudge, and music down just a nudge.

You also said:

The compressed sound will peak lower because that's what compression does, so if you try to match the previous level on a meter it's gonna be cranked a lot higher. Volume match by ear and then you can assess whether the compression is actually better or just tricking you with loudness.

I do this completely differently. First of all, in most cases, adding a compressor will increase the dynamic range of your sound. This is because initial transients will make it through before gain reduction kicks in. Once gain reduction kicks in the volume will be lowered, thus creating a larger volume gap between initial transient and the body of the sound. What I do to judge compression, I make sure my peak levels are the same before and after compression, and then judge the way the compressor is working. Peaks are my real nemesis. Big peaks make sounds small. (before the linear amongst you flip out, there is not a linear relationship there, there is a balance and a curve of efficacy). Let's say I'm pulling 4 db of gain reduction in the comp, and peaks are the same, if I after the fact add those 4db back to the sound in makeup gain, now I'm peaking 4 db higher! Not cool at all. That sound is going to have to be tiny to accommodate that big peak.

So maybe to simplify, I match my peaks before and after, and then judge what the compressor is doing. My aim is usually to get as big a sound out of as small a peak as possible.

Gotta run, but feel like I didn't draw the distinction I wanted. Oh well, time is short!

[SunkLo]

You need a limiter afterwards my friend. Or soft clipping, etc.

If you allow the compressor to squash the body down, you'll have a rough time A/Bing the compressor's effects. Your ear's not quick enough to perceive fast transients as volume, the body of the sound will be what affects your perception of loudness.

I've actually read of an approach where you let all those transients get through so there's still some dynamic material to trigger a compressor on the master or instrument buss. Then after you've got a limiter into a clipper to catch them all and round them off. Makes your master compressor more "accurate" in a sense because it's being triggered by sharp peaks instead of having to pull the threshold down into the body.

Wait till you get your hands on my compressor though. It compresses up and down so the average level converges towards the threshold instead of -inf dB. You still get to squish the dynamic range of the body without creating a big gap in volume between it and the transients.


Your fader approach is another valid one and quite common I think. Guys will go a step further and mix while tracking so when you pull all the audio stems up on the desk, it's already most of the way there with the faders zeroed. You're doing the virtual version of that I guess.

[nowaysj]

You need a limiter afterwards my friend. Or soft clipping, etc.

Tell me about it brudder. I am massively searching for a channel limiter that is transparent, 0 latency, and cpu light.

If anyone has recommendations. Please for the love of god post them up.

FabFilter's Pro-L is just an amazing limiter in my estimation - it is transparent (to my dumbo ears), but it introduces latency, and is not cpu light (when used across 20 or so tracks). I use it as a master buss limiter. Some times on busses, like if I have a drum buss and a music buss coming together, I might hit each of those with pro-l, before they hit the master. That way, the latency is equal between them.

[SunkLo]

Transparency implies lookahead which means latency. Does FL not have auto PDC?
You can just put a shit limiter on and then render out with Pro-Ls when you're done. Not the most practical but would give you a bit extra transparency on your final cuts.

Might also want to look into soft clipping, can be more transparent. Assuming your transients are short bursts of HF, the harmonics added by clipping will also be short and HF. Meaning the character of the peaks aren't dramatically changed apart from level, and there's no pump like a limiter.

[nowaysj]

Fruity does have apdc, but imo it is shit. Sometimes plugs don't report properly. Like if you are making adjustments within the plug that affect the plug's latency, it seems that this isn't compensated for by fl. Like when you instantiate the plug, the latency is read, and that is used for the apdc. If you change the latency, you're fucked. Or so it seems. I've been fucked by this. Fucked by behavior that isn't supposed to be happening. And fucked if I can figure it out. I'm still on an older fl, like 10 point something, but something low. I really should upgrade to 11, hoping that they did a better job.

Oh and also, it seems like the apdc only works to a certain buss depth. That is why I was asking you about reaper. I'm often several busses deep, and it seems like fl only compensates for like 2 or so busses deep.

As you can tell I really don't know what the hell is going on with pdc in fl, and don't really know how I can figure it out. I have a hard time on the forums over there. They're so edgy (like users and devs alike). Some of the support staff are aces but I try not to tax them too much

[SunkLo]

Oh word that sucks. There's a call in the VST API called SetLatency I think. FL must not be listening for that. They should at least have a manual button that re-polls all plugins and updates their latency. Maybe it's something they've fixed in the latest version. Check their changelogs/DL a demo. Isn't FL free upgrades for life or some shit? I'd def be trying to get on the latest version for issues like this. They won't take any bug reports seriously for older versions either so it'd still be a good idea to upgrade in hopes of getting them to fix it if it's not already.


As an alternative, you could always put a Pro-L on every track so they're all the same latency. Unless FL's PDC compensates for some and not others...

[nowaysj]

Yeah, I should update. It is one of those things - I'm working on tracks, I don't want to interrupt their progress, etc. I'm getting closer and closer to making stuff I'm happy with, I don't want to burn anything. I've got hundreds of song files from 8, hundreds if not a thousand+ from 9, etc, I don't want to leave behind thousands from 10

Also don't want to take the cpu hit of a new version. Each new version is more demanding, and I'm still rocking a system I built in 09. It was a good system, as evidenced by its longevity, and I did it for around a grand, but I don't have a grand to spend now.

And as a corollary - I don't think I can afford the cpu cycles of 20 pro-l's, on top of everything else. That is why if I use it on the buss level, if ever.

And I don't know if fl is not listening for SetLatency, or if some plugs aren't using it. I really should like kind of do a state of the union, and get to the bottom of this, but it is like go go go all the time. That's why building this damn drum sampler is such an utter ball ache, as it involves hours and hours of manual reading to get the simplest shit going. I'd rather be making a beat. Anyways. Enough of nwj's woes today!

[outbound]

I have FreeG at the start of each channel strip. It's useful if there's automation on the channel and I want to adjust things without messing up automation.

Mainly when dragging in samples it's useful to bring down the level before sending it to any plugins (these samples are usually normalised and any processing can clip the in's of any plugins put after it if level isn't brought down)

However if there are any dynamic processors in there (compressors, de-essers, expanders, freq-dependant EQ etc) then I will use FreeG after these so I don't have to reset the controls for all of these again (nightmare!)

[OfficialDAPT]

You should definitely use the Utility device if you plan on changing the volume of a track throughout a track. That way, you can still adjust the levels using the channel fader during the mixdown stage. If you are turning down the volume fader, you should still make sure that the signal is not clipping anywhere in the chain, even if it's not clipping on the channel master.

[AxeD]

Why would you not want it to 'clip' anywhere in the chain?
If it's even really clipping.

[SunkLo]

Because that's gonna fuck up the gain staging of any non linear plugins.