A few words on achieving mix depth with compressors

[DarkRed]

In this post I will share some pro knowledge on how to achieve mix depth with compressors.

The key to mix depth is in my view about staging compressors. What I mean by that is that the focus is not on the final brickwall peak limiter, the focus is rather about hitting the final brickwall peak limiter with optimal transients. These transients must be optimal not only in a single dimension but across dimensions. What I mean by that is that the dynamics should be great no matter whether it is on the individual sound source level, whether it is on the group level, whether it is on mid the component, or on the side component, in the verses, in the choruses... All dynamic perspectives should be in balance when they hit the final brickwall peak limiter.

In my mixes I typically have these dynamic stages:

A) Sound source signal leveling

B) Two stage sound source compression (side-chained)

C) Low, mid, high compression (side-chained)
<--- STEREO WIDENING --->
<--- EQ --->
<--- TAPE SATURATION --->
<--- EQ --->
C) Two stage mix bus compression
C) Mix bus compression
<--- TAPE SATURATION --->
<--- EQ --->

D) Multi-band compression
<--- EQ --->
D) Mix bus limiting
<--- EQ --->

So prior to the final limiting the content has passed 6 stages of signal leveling (excl. EQ and saturation).

Now, in order to create depth and size, the technique I use is to mix into the (C) compressor stages above when I dial in the (B) compression stage. This means I mix into a particular sound stage that allows me to set the dynamic balance of the mix elements to where the sound stage sounds the best, but on the individual sound source level. For width I pan into stereo widening.

When I go for a pop or country mix I want a low attack foot print on the mix, meaning that all sound sources when they combine should have low peak-to-rms ratios present inside of the audio, so that even at commercial loudness levels the mix sounds soft. To do this, the earlier the compression stage the harder I want it to act on the content, meaning that ideally I want to be able to dial in the (D) compressor stages gently by focusing most of the compression on the (B and C) stages.

This means that on all sound sources present in the mix, to begin with I first of all ride the incoming signal a little (stage A), so that all peaks on all tracks stay within a certain dynamic range that limits the combined peak-to-rms ratio down to a default level that makes it possible to end up within a certain final mix attack foot print range. These days that range is pretty extreme, so the initial signal leveling is hence very important. This you can see visibly in the form of a "thick" mix waveform.

Then during the mixing process I enable the compressors on the C stage and dial in the compressors on the B stage using two stage compression.

Then during the mastering process I dial in the compressors on the D stage according to the optimal dynamic profile I want to achieve, this also includes expanding frequencies. At that point when I have ended up with optimal dynamics across the frequency spectrum, I fine tune with an EQ effect after it. Then it is that final signal that I maximize with a brickwall peak limiter.

So in other words I add compression in stages. I set the signal level on each track to focus the mix towards a certain peak-to-rms potential.

I control the overall dynamics by distributing the compression duties across stages. I mix into compression to utilize the full potential of the sound stage and in order to get optimal compression ratios on the individual sound sources. I add dynamic breath using two stage compression, on the individual sound source scope and on the mix scope. I work with the mix dynamics separately from the final dynamic profiling by making the signal that hits the D stage compressors near the kinds of dynamic profiles I am aiming for. Finally I am applying the desired dynamic profile (typically by testing out a few various ones), tune the resulting frequencies to achieve the mix frequency balance that I want and finally I maximize the signal up to the loudness level I want, after that I might have an EQ that might do some final adjustments, depending on whether I find I need it or not, but often I want it because I want the final version to have a specific frequency response on the very detailed level at the particular loudness. The application of the D stages results in various prints and I finally pick the print I like the most. In more demanding projects I make the mix go through several dynamic profiles, because I might for instance find the dynamic profile of the verses be optimal in print 1 and the dynamic profile of the choruses to be optimal in print 2. So the resulting wav file could have a complex set of dynamics processing that has gone through a rigid selection process. During mastering I am very focused on what my ears like, so whatever they like I go for. This means that the final version might have very specific comp/EQ combos on various song sections.

The reason why I have quite a lot of EQs is because I want to maintain a good overall frequency balance through the processing stages (have a good input to the next stage) so that I don't overdo the effects and to improve the performance of the effects before it.

Furthermore, the vocals I side chain to the kick-bass-snare combo, this means I can reduce the volume of the vocals (even more by lo-pass filtering the snare) and in return get a more clean and full low end. (because I hi-pass the vocals but the side chain allows me to hi-pass a bit more gently) I also side chain the kick and bass, so that I can lower the mix attack coming from the kick. The snare clearity I typically achieve from the low mid high comp side chain in combination with the other side chains and because I scoop out a little of the mids of the bass.

The density/warmth of the mix I tend to set using hi-pass filtering on the sub frequencies of the mid and side components separately prior to all of the saturators/compressors in the C stage.

 

[Synapsis]

Your talking analogue chains? Not digital shit?

 

[DarkRed]

Yes most of these stages should be carried out with hardware, especially the final brickwall peak limiting.

It's similar to the difference between a real B3 organ with a real leslie speaker cab, versus a software B3 with a software leslie speaker, it sounds similar, the difference is in the amount of depth, the real B3 organ with the real leslie speaker cab has more depth. When you do it with software the effect is similar sounding, how similar depends on the specific software, its settings and the application of the effect. When you do it with software the best bet is to use the plugins that are cutting edge and model a specific type of effect, like for instance a FET compressor, then try to come as close as possible through the way that is being applied.

The performance of hardware vs software depends on a number of factors. For instance if you apply hardware effects in the wrong way using poor recording and monitoring quality and compare that to using the right software effects in the right way using great recording and monitoring quality and great DAW performance, the software might actually win. But if you are using the right hardware effects in the right way using great recording and monitoring quality, that will win. Keep in mind that software is much cheaper though, so since the budget is also a limiting factor on your setup as a whole, you need to decide what gear that needs to be hardware and what gear to make tradeoffs for by using software.

Also keep in mind that with hardware you can do things that very easily turn into phase issues with software, processing techniques that contribute with depth. For instance two stage compression on individual sound sources with software will demand the same compressors to be used with very similar settings, because else the tracks are going to start shifting in time relative to each other. (because their delay is essentially multiplied by the processing behind them) And if you want to blend the processed tracks with the dry tracks, then that can often result in a delay between the two that will create phase issues. So there are things you can do with hardware that are difficult to do with software with the same result, because with software you also have latency. How much will totally depend on the performance of the DAW, but for most it's an issue.

With software it can pay off to mix in stages and do prints when each stage is done in order to keep the latency at a minimum, to keep offloading the DAW while at the same time un-restriciting the amount of processing that you can do on the mix content. Because interestingly with software you often need to overdo in order to get a near similar amount of character as with hardware and that specific overdoing creates issues ITB.

If you for creative reasons dynamically level instrument tracks with several volume faders per sound source throughout the mix, that process is a typical stage you want to do in isolation and then print in order keep the phase between those at a minimum.

Overall I think it's great to separate the work you do individual sound source oriented, in its stages and the work you do group oriented, in its stages. If you for instance have a mixing model where you apply comp side chains only on some frequency bands, lets say between the low end of the kick drum and the bass, then that's a kind of work I would want to do in a particular point in time in the mixing process and then print before resuming the rest of the mixing. Ultimately the staging of the mix allows you to keep the number of faders and fxs to work on at a minimum, which makes you more effective and more focused on the mix. If you do it this way you can also always keep overwriting the dry tracks to sync the delay times across the prints. Then on the last stage before resuming to the group level stages, you balance the wet to the dry ratio of the sound sources in order to get the most optimal final transients and characteristics and then print those. (if you do it the right way it becomes two different colors of each sound source that you blend by balancing them (maybe you want one of the versions on each side to stereo enhance them) or by choosing either one of the versions, this is a creative process) Now you have bypassed the issues that otherwise would come by doing it another way ITB.

 

[chris carter]

I think that was pretty confusing for most. I think it would be great if you could provide an example of one of your mixes where you use this technique so people can hear what you are talking about. That would make it easier to understand.

 

[DarkRed]

I think that was pretty confusing for most. I think it would be great if you could provide an example of one of your mixes where you use this technique so people can hear what you are talking about. That would make it easier to understand.

It is confusing because it is advanced and I have left out things in it that would make it even more advanced, things like memory leaks in the audio drivers on the Windows platform, why resetting the CPU affinity should be performed prior to a print, why the dithering should be bypassed through external routing, how oversampling impacts on things and stuff like that.

 

[MarcDHall]

So... use a bunch of stuff and get things right before hitting the limiter.

Gotcha, thanks.