Mastering

The magic frequency in mastering#

Which frequency should you always boost in mastering? Internet people are liable to answer that question with a number. An AI chatbot told me to add 1 dB at 10 kHz for clarity in the highs. Obviously that is stupid advice.

The correct answer is of course that the question is stupid. There is no conveyor belt process. People picture a mastering machine: the song goes in and comes out "masterised". That is not what mastering is. EQ, compression and limiting are just tools. The main thing you need is good monitoring and experienced ears.

The mastering engineer makes the master disc#

The term "mastering engineer" comes from "he who cuts the master disc", not some kind of "master of the audio realm", some gatekeeper wearing a superhero cape. A mastering engineer used to have a reasonably skilled technical job - not blowing up the cutter head with excessive HF content, not causing excessive vertical modulation that could make the stylus skip.

Today, because most of us don't work lathes, the mastering engineer has simply become the last point of call before release: so essentially a quality control checkpoint and tonal balance specialist for when something is unintentionally tonally skewed (hence why tilt shelves are important and a core feature of Dogma). Stuff like "emotion" is supposed to already be a wrap in the mix stage. If it's got no emotion we're probably all wasting our time working on it. It is not my place to arbitrarily start putting flanger effects on stuff, but appears there are many people out there who think that their role is to run stuff through old metal boxes to "give it mojo" or whatever. That's not your job. Vent your supressed creativity by making your own music - don't use other people's art as an oppertunity to think you are doing anything creatively meaningful. The role of a mastering engineer is technical only.

I learned this early in my career. Someone sent me a song to stem master with badly played acoustic drums. I deleted his drums, jumped into the live room and bashed out a significantly better drum recording (back when I was actual practicing every day), and then sent him back an absolutely great sounding version of his song. I thought he'd love it. I thought I was doing him a massive favour. He sent me something which was "just OK" and I sent him back something which sounded great. And all of this for the price of normal stem mastering. What a deal he was getting!

And I never heard back from him again.

Why? Because it was an insult. "Hey, your drums were crap so I did a better job myself". I was so motivated to do the best job, I didn't realise that part of doing a good job included respecting the art of the artist. I took his art and pissed on it.

Plausibly the biggest mistake of my career.

I'm telling you this because mistakes are to be learned from. And the "I'm going to run your music through these old metal boxes to give it mojo" is a somewhat more mundane but equivelently unpleasant manifestation of the same underlying piss.

The job is to listen, assess and talk#

The correct approach to mastering is for an experienced, highly competent audio engineer, knowledgeable about the genre and instrumentation of the music, able to appreciate its composition and aesthetics, to listen. To assess. To create dialogue around the mix and presentation. To work with the artist or mix engineer to achieve the greatest possible result. And only after this process of critical examination, quality control and understanding is satisfied may some additional tonality and loudnes processing be performed to create the master files for digital distribution, vinyl cut, tape or CD.

Usually, what a record needs is not more processing on top of the mix. It's mix feedback, asking for a specific stem or understanding the broader context surrounding the track on an album. If you want to add random additional processing to your mix, AI is pretty good at that now. Experience, mix critique and judgement are the things it cannot do.

A master is the product of the whole chain#

A great master needs the tonality right, which needs a good mix, which needs a good recording, good musicians and a good composition. Itsy Bitsy Spider on a glockenspiel will not come out of mastering as a wall of symphonic metal.

So a good mastering engineer has worked in studios, played in bands, recorded and composed. One who only knows how to run things through a limiter is crap and the robots will take his job.

Stem mastering is not a separate service#

"So how much does stem mastering cost?"

This is conveyor belt thinking. Stereo mastering is service A, stem mastering is service B, vinyl masters are service C, and an analogue EQ instead of a digital one is some other alternative service, as if I were a barista and upgrading your cappuccino to oat milk costs an additional euro. There is no price list for Espresso, Cortado, Decaf or an additional squirt of hazelnut syrup in my approach.

I will listen, and I will hear whether it needs stems or not. I will notice if the mix has issues, and discuss them. In many cases that will be: OK, sounds good. In many cases it will be: can you send over the kick stem? I work on the stereo sum and only start looking at stems when I hear a problem.

Limited and clipped mixes#

Headroom is trivial; everyone has a volume knob, so aiming for -6dbFS, -10dbFS, -1dbFS is all the same and doesn't matter. What matters is whether the mix has been limited or clipped, which bakes pumping and distortion in for good, like firing clay in the kiln and then asking someone to mould it. Pulling bus processing off can substantially change a mix, in which case I'd suggest sending the unprocessed mix plus the demo master as a reference.

Version 7 has never been better than version 1#

I do not count peas. I do whatever is needed to get the music finished, and sometimes what is needed is pushback against endless revisions. Version 7 has never been better than version 1 in all of my career, not when it comes to mastering. It simply demonstrates the music was not ready for mastering: the artist was not yet content with the music or mix, which is fully outside the scope of mastering. I do not do version 7, nor 6, 5 or 4. Either the artist loves the initial master, perhaps with one small tweak, or I am not the right engineer for the project.

Secret gear#

But what gear do you use? If you feel the need to ask, the answer is simple: I have a collection of audiophile Gameboys with gold batteries and gold audio cables, and I process everything using 22 MHz sampling rate DSD through nine quadraphonic flanger instances. This is the secret to my sound. Unparalleled clarity, every single time.

The real answer is boring. I've found digital mastering gets better results than analogue mastering, so I sold my analogue mastering gear. Flicking between tracks with instant recall beats running one track at a time through old metal boxes. Mixes already arrive saturated and compressed, so adding more analogue circuitry for harmonics is almost always an amateur mistake. Mastering can be as simple as raising the input into a well-configured limiter until it sounds right at the reference level.

This is why I created Dogma and why I use it exclusively for every master I work on. I do not use any thing analogue because using anything other than using Dogma gets worse results.