IMO the idea that an audio system, for recording or replay, needs a dynamic range of 123dB is moot.
It would only be necessary if you wanted to make a recording of
any sound without adjusting the level control, ie record a trickling stream out in the country and then show up at a rock concert with the same recordewr and microphone and wish to record that too without adjusting the input level. Bonkers.
Same listening at home assuming you never touch the volume control, want even the lowest levels on a recording to be audible and are prepared for permanent hearing damage when a loud bit comes along. Equally bonkers in practice.
In reality, if I am recording a trickling stream out in the country, I will be turning up the levels on my recorder and my limiting factor monitoring the levels will be listening for microphone or other noise in the quiet bits, and the dynamic range - from the loudest trickle to the quietest inter-trickle moment will easily be precisely recorded within 16 bits, never mind 24.
Similarly when I bimble out of the country into a rock concert with my trusty recorder and microphone, when the band is warming up I twiddle the level control - probably a long way down
and then, when monitoring (theoretically, headphones won't isolate from the racket enough to monitor levels when in the concert itself) I will be making sure I don't clip. Yet again the difference in loudness between to loudest bits of music and the witty between song dialogue will be easily covered by 16 bits of dynamic range, never mind 24.
So in real sound recordings 16 bit is plenty.
24 bit has the theoretical (and pointless IMHO) benefit of recording every sound there is without adjusting volume - but who in the real world would ever think of doing such a thing?
The difference between theory and practice...