On Dogma

The other day I taught a small-group lesson to a few of my college students. One (very talented) student brought in a bounce of a song he’d produced and mixed for a friend of his.

It’s really quite good- great performances, compelling songwriting, a quirky sparse arrangement that leaves a lot of room for serious width in the stereo field. Great recording and sound design, punchy drums. It’s music I’d legitimately want to listen to.

But all his hard work and skill was being significantly counteracted by the inadvertent downsides of the processing on his master fader.

The final limiter was doing a good bit of gain reduction, and set to True Peak, and with a *long* lookahead. The EQ beforehand had the usual generic HPF at 30Hz and LPF at 18kHz or so. There was some saturator in the chain providing mostly even-harmonic distortion.

And none of that was helpful to the sound of the record or the feel of the music.

So, one step at a time, we talked through his reasoning for each processor in the chain, and we found out (as I expected) that a good bit of the master fader decision making was out of habit, or was following dogma from various internet audio sources.

We talked about the limiter settings, and why even the fastest lookahead setting available on this particular limiter (20ms) was too long for the music in question. The production had some nice, overdriven, crunchy, powerful drums. The 20ms lookahead does a great job at avoiding distortion, but this is a production that’s already embracing distortion. The lookahead time plus the True Peak detection was seriously pushing the drums to the back, even with only moderate gain reduction. That was obvious to my ears, because I know to look for it. It wasn’t yet obvious to my student’s ears, because he hadn’t yet thought about it. That’s not his fault of course- we’re all learning, and there’s no way to know until you know, and in any case he’s years ahead of where I was at that age. But, he was largely following the (in his defense- a popular) mantra that overly severe limiter settings and inter sample peaks can cause problematic distortion. And of course that’s true, but also of course it’s not the whole story.

So instead we spoke about the notion of positive compromise: The idea that most any audio processing choice has both benefits and downsides. And that it’s usually very easy to hear those benefits, and it takes a good bit longer to learn to hear the downsides. And that ultimately, you have to weigh the trade-offs and find the path forward that best serves the music.

We used a different limiter, one that allows behavior much closer to a clipping shape: Short lookahead times and very fast release. And we turned the True Peak detection off. Immediately, the drums came to the front, there was a marked increase in transient content, loudness, and width. The only downside is an increase in distortion, but it was not at all noticeable, and the music accommodated the processing just fine.

We then talked about his HPF and LPF decisions. Again, those were not choices that had been made based on anything audible, just “best practice” (not really actual best practice….) as per the recommendations on the internet.

So we talked about the phase implications of EQ. How even a 30Hz HPF can rotate phase all the way up into the midrange, with a corresponding increase in peak level, especially when there’s significant distortion in the existing tracks. And we talked about how that can push any downstream dynamics processing that much harder.

I showed him the test case of a 100Hz square wave with a 30Hz HPF applied, and how that bumps the peak level up by 8dB, without any increase in perceived loudness. That’s obviously a worst-case scenario, designed to maximize the impact on peak level, but his actual song had enough odd harmonic distortion where just turning off the filters reduced the peak level by about 2dB.

There was also a subtle increase in width from removing the LPF, and again that aspect worked very well with the music.

I firmly believe that the technical execution of a certain audio task is rarely difficult in and of itself. It takes all of two minutes to show someone what a high shelf looks like and how to turn the knob on the EQ. It takes a similarly short time to sweep a compressor attack time between fast and slow, and demonstrate the resulting difference. And usually, in isolation, any number of approaches could theoretically work just fine. 

Instead, the sophisticated aspect of the craft comes from understanding the music and sound in front of you, and understanding how you yourself and the rest of the production team feel about that music. And then comparing sound + feeling to the *intended* sound + feeling, and figuring how to move things closer to that intended state. 

(And, all the while, keeping in mind the history of recorded music as a yardstick for calibration….)

On the one hand, that approach can be a bit scary, because it means that a wide range of often quite divergent processing decisions are all potentially viable. It means you might use some tools in some ways on one mix, and then tackle the next mix in a largely opposite way. And that can certainly make the work harder at first. But on the other hand it’s ultimately liberating. Once you put in the work and get to the proverbial other side, it leads to a place where you are attuned more deeply to the music in front of you, and you’re more likely to make decisions in service of that music and the overall artistic vision.