In contrast to production techniques applied to popular music, most published recordings of classical music can have a very wide dynamic range. This makes then less suited for listening in a car, on phone line, or other contexts where there is a lot of noise: if you were to boost the volume of your amplifier so that you could hear the quiet parts, you would get blasted when the dynamics change in the music.
When listening to my local classical radio station, I believe they are effecting the source recording to make it more appropriate for radio and listening in a car. I can recognize a number of things by listening to the noise floor of the track; namely, that when the music is very quiet (like a solo instrument in an orchestral piece), the noise floor is louder, indicating that the track has been boosted. But also, that when the full orchestra is about to loudly come in, the noise floor drops out just before the increase in dynamics actually hits, indicating that the boost has been removed.
How is this being accomplished? Is this some kind of compressor, or a predictive volume normalization algorithm of some sort, or is there just some engineer with their finger on the fader who knows the piece well enough to adjust the volume manually?
I would like to know more about this so that I can replicate the same technique on a collection of archival recordings of classical music for use as hold music for a phone system.