Quote:
Originally Posted by
darklide
➡️
I see the new Motu M2 does this interface record in 32 bit float or just 24 bit?
What do you expect by "record in 32 bit float"?
Anything that goes into a computer will probably be 32-bit float pretty quickly—certainly anything that goes through Core Audio on Mac OS X is piped in 32-bit floats. There is no degradation, because 24-bit map into 32-bit float exactly (with a bit to spare—32-bit float with normalization and size bit gives the equivalent of 25 bits of mantissa—the remains bits are for the offset exponent). It doesn't matter whether the box did the conversion or the computer, nothing really changed.
But electronics can't even resolve 24 bits, not even close, so there is not sense talking about filling that 25th mantissa bit or getting more precision at less than full scale with the exponent.
No, really, it's not possible except near absolute zero (and good lucky listening on speakers at the temperature. Besides, you'd be dead. And your ears can't resolve it anyway, even if you weren't dead).
So, why have 24-bit converters? Because we wanted more than 16, and we use computers so we added another 8-bit chunk.
Found this just recently—and mind you, this is written by a company that makes a $7,555.55 DAC, and their cheapest one is $2.5k:
"Any company that claims greater than 20-bit resolution from their DAC is simply full of ****." (from the article "The 24-bit Delusion").
Some manufacturers include terms like "32-bit", but it's always some meaningless marketing ploy that references some intermediate precision in their oversampling filter, etc. Basically they're lying, with an out that a component has a 32-bit somewhere in its spec sheet. Don't pay extra for anything because "32".