They didn't calibrate it themselves???
The starting point would be the Fletcher-Munson curves:
But nothing short of meta information provided with the recording can take into account the 'correct' level for playing back the recording.
If the recording is of a quiet 'event' then even though it is quiet, you *shouldn't* boost the bass and treble when listening to it at a realistic (yet non-neighbour upsetting) volume. Only if the recording is of something loud and you are playing it quietly should you boost the bass and treble.
But the volume control setting can't help you. If the recording features just an acoustic guitar, they are not going to metaphorically cut it on the metaphorical disc at the same gain as a full blown symphony - it will be 'cut' at a different gain.
Nor can the recording's content be automatically analysed to tell you whether to boost bass and treble for a particular recording unless your algorithm can deduce the absolute 'correct' volume from content. I presume it is also accepted that you don't want to boost bass and treble
dynamically with content - yuk!
Unless a recording is supplied with information about the absolute level it should be played at, there is no scientific way to do it. Without this information there is no "calibration" possible. It must always remain a subjective guess.
And even if such information was supplied, how would they take into account all the deviations from 'raw', including dynamic compression that is changing the recording's position on the Fletcher Munson curves all the time? (one of the reasons why dynamic compression is evil!).