How to master against a reference track

Every producer already masters by reference — it's just usually a vague memory of how the last record they loved sounded in the car. Doing it deliberately, with the reference actually loaded next to your track, is the single fastest way to make mastering decisions that survive contact with other people's speakers.

Why referencing works

Your ears adapt. After twenty minutes inside your own track on your own monitors, "bright" and "bassy" stop meaning anything absolute — your brain has normalized to whatever it's hearing. A commercial reference resets the scale. It's a known-good tonal balance, mastered by someone with better monitoring than most of us, already proven on club systems, radio, and phones. When your track and the reference sit side by side at the same loudness, the differences you hear are real information, not fatigue.

Choosing the reference

The reference does most of the work, so pick it with intent:

• Same genre, same energy. A melodic techno track referenced against big-room festival masters will come out overhyped in the top end and crushed. Match the subgenre and roughly the arrangement density — a sparse dub techno cut and a wall-of-sound trance anthem have legitimately different spectra.
• A master you'd sign. Recent, professionally released, and something that already sounds right on systems you know. If it sounds harsh in your car, it will drag your track the same direction.
• Lossless source. Use a WAV/FLAC from a download store, not a rip. Lossy artifacts pollute the top octave of any spectral comparison.
• One, maybe two. Averaging five references averages away the point. One primary reference for balance, optionally a second to sanity-check the low end.

What to actually match

Four measurable dimensions cover most of what "sounds like a record" means:

1. Tonal balance. The long-term spectrum — how energy distributes across the bands, typically viewed in third-octave resolution. This is the biggest audible difference between an amateur and commercial master, and it's very learnable: pull up an analyzer on both tracks and read where the curves diverge. Persistent gaps of 2–3 dB in a band are worth an EQ move; chasing every half-dB wiggle is noise.
2. Loudness. Integrated LUFS for the overall level and short-term LUFS in the drop. If the reference sits at −7 LUFS and you're at −11, your track will feel small in any un-normalized context — see how loud should my master be for choosing the target honestly.
3. Stereo width. Compare the sense of space band by band, not overall: commercial dance masters keep the low end essentially mono and open up width through the mids and highs. A correlation meter plus your ears in mono tells you most of it.
4. Dynamics. How much the track breathes between kick hits. Crest factor and short-term-vs-integrated LUFS spread are decent proxies. If the reference pumps musically and yours is a flat wall, the difference is usually mix-bus and limiter behavior, not EQ.

The level-match trap

One rule makes or breaks all of this: compare at matched loudness. Louder is reliably perceived as better — fuller, wider, punchier — even at differences well under 1 dB. If your reference plays 4 dB hotter than your premaster, every A/B "proves" your track is worse, and you'll EQ and limit toward a mirage. Turn the reference down (never your track up into clipping) until both sit at the same short-term loudness in comparable sections, then judge. Every serious referencing tool level-matches automatically for exactly this reason.

Doing the match manually

With differences noted, work in this order: broad corrective EQ first (wide bells, small gains — you're steering a finished mix, not remixing it), then dynamics if the density differs, then stereo adjustments, then limiting to the loudness target with a true-peak ceiling of −1 dBTP. Re-check against the reference after every stage at matched loudness — moves interact, and it's common for a 2 dB low-mid cut to make the master suddenly read brighter than the reference.

Know the limits, too. Reference matching shapes what mastering can shape: spectrum, level, width, density. It cannot move a buried vocal, fix a clashing bassline, or add arrangement energy that isn't there. If matching the curve still doesn't get you the sound, the gap is in the mix — often the low end — and the fix belongs upstream, in a properly prepared premaster.

Automated reference matching

The manual process above is measurable end to end, which is why it automates well. Matchering-style tools fingerprint the reference's spectral balance, loudness, and width, then compute the EQ and dynamics moves that carry your track onto that curve — the same third-octave comparison, done in seconds without ear fatigue. The good ones show you every move they made so you can veto them. That's exactly how Sunset's reference matching works: drop in the track you want to sound like, and it reads the balance, loudness, and width, shapes your mix toward it band by band, and reports each decision in plain English.