On "Yanholom," the lyrical single Kalush Orchestra released in February 2024, the label metadata collected on AllMusic lists me four times: composer, producer, mixing engineer, mastering engineer. Same picture on "Stefania," where Sony's release metadata on Qobuz carries all four roles. So my answer to whether you can master your own production is yes, I do it on commercial releases. The honest part of the answer is that the workflow only holds up because specific checks replace the thing it removes, which is a second pair of ears that has never heard the song.
What disappears when one person holds all four credits?
Handoffs. In a conventional chain the producer catches the writing problems, the mix engineer catches the production problems, and the mastering engineer catches the mix problems, each of them hearing the song for the first time with nothing invested in the decisions already made. I walked through where those jobs begin and end in what separates a producer, a mixing engineer and a mastering engineer. Collapse the chain into one person and every one of those fresh first listens is gone. By mastering day I have heard a song like "Yanholom" some two hundred times, and the ear stops reporting what is there and starts reporting what it expects. Adaptation is the real risk, and it grows with every playback.
Why is the quiet single the hardest case?
"Yanholom" is a slow, open love song from a group the public met through the Eurovision win, and quiet arrangements punish self-mastering more than dense ones. A wall of synths and drums masks a 2 dB imbalance; a voice over sparse accompaniment hides nothing, and the temptation the producer side of my brain whispers is to push the limiter until the ballad sits as loud as the bangers in the playlist. Platform normalization makes that trade worthless, the arithmetic is in what I send out of a mastering session, so the discipline on a track like this is the opposite: keep the verses low, let the last chorus carry the lift, and accept an integrated loudness the loudness war would have laughed at. The composer in the room wants the song protected. The mastering engineer has to want the same thing with meters.
What replaces the fresh ears?
A protocol, written down, followed every time. Mine has four parts.
The gap. The mix gets printed, and the session closes for 24 to 48 hours. No listening in between, not even in the car. The master pass starts cold the way an outside engineer would start it.
Level-matched references. Three commercial tracks in the session, loudness-matched to my printed master within 0.5 dB, switched blind. Unmatched referencing is self-flattery, because louder reads as better for the first thirty seconds. The references are chosen before the mix is finished, not after, so they judge the master instead of excusing it.
Different rooms, different speakers. The same master plays on the mains, on a deliberately mediocre second set, in mono on a single speaker, and in the car. The car run has its own ritual, documented in the Van Nuys Boulevard car test, and it stays the most reliable lie detector I own: a low-mid buildup my treated room forgives shows up on a freeway at 65 mph inside one chorus.
The edges and the numbers. Every rendered file, top to tail, tails and fades checked at full attention, then true peak confirmed at -1 dBTP, sample rate and bit depth confirmed against the sheet, the dithered 16-bit version nulled against memory at matched volume. Boring, mechanical, and the step that catches the export region set a half second short.
Does the collapsed workflow have advantages?
Two real ones. Intent survives: nobody has to decode what the low end was supposed to do, because the person who arranged the bass is the person mastering it, and on "Yanholom" the space around the voice is a writing decision as much as a mix decision. And revision cost drops to zero: when the master reveals that the second verse needs the pad 1 dB lower, I reopen the mix and fix the cause instead of bending the symptom with mastering EQ. An outside engineer can only process the stereo file in front of them. I can go back one layer. That round trip, master to mix and back, is where a one-person chain genuinely beats the conventional one, as long as it runs on the protocol above rather than on confidence.
When do I still send work out?
When the project needs a check I cannot give it. A record going to a label, a style I touch twice a year, an album finished in a sprint where the gap days do not exist. The deciding question: can this release afford my blind spots on this schedule. Cross-genre compilations are the clearest case: a contest season produces stems and versions in bulk under hard deadlines, and the delivery discipline matters more than authorship pride. The same logic runs the other way for other people's projects, where I am the outside ears being hired.
What to do with this
If you produce and mix your own material, run one experiment on the next single: print the mix, stay away from it for two full days, then master it against three level-matched references on two playback systems plus the car. Keep notes on what the gap revealed. If the list is short, you are closer to self-mastering than you think; if the list is long, the notes are the brief you hand the mastering engineer you hire. The releases this workflow comes from are documented with sources on my credits page, and the longer background is on the about page.