Public reference

The measurement archive

Most of the numbers we master against are not published by the platforms themselves. When one of them moves, you should be able to see when we noticed and what it was before. So every check is kept with its date instead of overwriting the last one.

Each check is one JSON file, linked below. Quote it, diff it, hold us to it.

Checks kept
1
Changes recorded
0
First check
2026-08-07
Latest check
2026-08-07

The changelog

What moved, and when

Nothing yet. There is one check in this archive, taken on 2026-08-07. A changelog needs two, so there is nothing here to read. That is the honest state of it: this page earns its keep on the day the second check lands and something turns out to have moved.

Every check

One file per check, kept as it was

A check that finds nothing is still a check. Without it, an old figure is indistinguishable from one nobody has looked at in a year.

DatePlatformsResultRaw
2026-08-076first check, nothing to compare2026-08-07.json

Why keep it

A number without a date is a claim

Most of these figures are not published
Of the 6 platforms in the table, 1 publishes its own loudness figure. The rest are industry consensus, which is another way of saying that everyone is copying everyone and the original measurement is nowhere. An archive is the only thing that makes such a number correctable instead of permanent.
The engine masters against exactly this
Each file here is written straight out of the same table the tool uses, so what you download is not a description of the tool's behaviour but the input to it. If a figure in the latest file is wrong, every master we made since that date inherited the error, and you can see which ones.
History does not get rewritten
A check is stored under its own date and never edited afterwards. Changing a value means recording a new check, not correcting an old one — including when the old one turns out to have been wrong. A quietly corrected archive would be worth less than no archive.