The Sound Quality Reading
CORPUS is being built in the open. Some of what you read here is live, some is still design intent. Expect it to evolve.
Every contribution receives a reading of its mix and master: how the sound sits against well-made records of its own style. The reading is advisory and awards no points. It answers one question: how far is this production from the reference practice of its own style?
Why the reading awards no points
There is no universal scale of good production. A great punk record and a great pop master are close to technical opposites, and a crushed, maximally loud master is a defect on an audiophile jazz record but the whole point of a club track. So every measurement is compared against a curated pool of well-made records of the track's own kind, never against one ideal. Which pool is the right one depends on what you were aiming for, and that intention is yours: no measurement can know it. A deliberately raw mix reads far from polished reference practice and is still exactly what it should be. The reading therefore describes and never scores. It reports difference, not quality, and leaves the judgment where it belongs, with you.
The reference library
The yardstick is a curated library of reference-grade studio recordings, currently more than 1,300 tracks across dozens of style pools, and growing. Every track earned its slot for a documented reason; many are the records working mastering engineers keep on the desk. Which pool your track is compared against comes from its annotations (genre, character, instruments). Accurate annotations pay off here: they make sure your track meets the right yardstick. Era is a style label, not a discount: a track that reads as 70s soul keeps the label but is measured against current production standards.
What gets measured
Mostly classical, readable descriptors a mastering engineer could argue with: loudness behavior, dynamics, spectral balance, stereo image, source quality, transient punch, how steadily the top end holds, how cleanly the two channels sum to mono, and, when the track has vocals, how the voice sits against the band. Two learned readings join them, always relative to the style pool: a perceptual production-quality estimate and a mix-style similarity.
Faults have a direction: a mix that is more dynamic than its reference pool, or breathes more in loudness, is never treated like one that is crushed flat. Healthy dynamics are not a defect here.
What you get
The reading names the style pool your track was measured against and the closest alternatives, shows how the sound sits across the production dimensions, and describes the strongest deviations in plain language, each anchored to the passage where it is clearest. If the named style is not what you were aiming for, you pick your intended style and the whole reading updates. Every statement names the measurement that earned it, next to the reference practice it was compared against.
Next: Novelty Estimation.