Hi Elias,
Thank you for jumping in.
Peter, operating at 24-bits has several advantages, even if the audio is 16-bits. For instance, operating the software's volume control will result in a greater-than-16-bits digital audio stream (because the math leaves remainders).
Maybe not to discuss this, and merely to put it in the context of us audiophools hence everybody knows what he/she can head for :
I don't think anyone with great care for his audio payback would want or should use software (aka digital) volume control. Note the "great care" though, of which I know that the people in here (say, this community) are about that to the highest degree. "They" hear everything, and so to speak, an i++ sounds different than an i = i + 1.
Now, unless you have a very dedicated / tuned set of amps allowing for 24dB attenuation max, 24 bits is not enough. Even 32 bits would allow for 48dB attenuation only without any losses, but, that would suffice with not too crazily over-powered amps. For me anyway (33W into 114dB @8Ohms and always playing loud).
So, using 24 bits with 16 bit data for digital volume control, honestly is not an argument IMHO.
Besides playing (original) 24 bit data - which seems "useful" enough to me to need 24 bit support in XXHighEnd - things can get rather complex;
First I'd say that using a 24bit DAC with 16 bit data, this DAC theoretically will be more (Voltage) stable at the 16'th bit. So, an advantage.
Next, however, I would be using a sigma/delta operating DAC then, which is far from my personal likings (and principles around XX) : non-oversampling.
Third, the first might not be true because of the 1-bit operating, me now not knowing whether there's advantage/disadvantage and if so, where.
Elias, I only responded to your implied "24 bits is better".
If that would be a ladder DAC (and for me without filters), yes.
Regards,
Peter