Originally Posted by jameskollar
I'm trying to make sure I have a fairly good understanding of 24 bitdepth vs 16 bitdepth as it applies to playback
. I understand that there are varying viewpoints on this but I would like to know if I am in the ball park with the following.
BTW: For arguments sake, sampling rate doen't matter but let's set it at 48khz
1) 24 bits is really only needed for recording and mostly because it allows for for more headroom in mixing. In other words, one an correct for deficiencies (i.e recording at an improper level such as -20db). I would think there are other reasons but this is not where I am going (yet) and I don't know what they would be.
Yes. And yes, there are other reasons besides correcting mix levels, such as insuring sufficient margin for accumulated signal processing stages.
2) 24 bits is has a theoretical dynamic range well above 140db which is well above human hearing.
And it's also beyond the limits of analog circuitry to reproduce--thermal noise.
3) A more realistic bit depth is 20 bits. The other 4 bits are really just noise. It would be difficult to tell the difference in either.
Not difficult, but impossible, all else being equal.
4) 24 bits properly dithered to 16 bits produces a signal that is aurally closer to 18-19 bits.
It is true that dithering allows the carriage of signals much smaller than the LSB. Let ma also add that we should never entertain any digital audio system that lacks proper dither, regardless of the wordlength. It's improper usage of the medium.
Following this logic, I would say it is posssible to take the end result of a fully mixed 24bit sound track, dither it to 16 bits, and play it back with very little loss (as applied to human hearing). Obviously you can measure the difference but I contend that all but the most discerning ears would not hear a difference especially as it applies to real sounds, not test tones. Apply the 5 channels to the mix, and I would say it would be very difficult to correctly pick out the 16bit tracks from the 24 bit tracks over consumer grade and perhaps even professional grade equipment.
I agree with your assessment.
Does this hold water? I know not all insiders would agree with this but I assume this is still open to debate.
I would only add that delivered 16-bit sources usually encounter further post-processing after delivery, and even if that is done in a 24+ bit environment, any slight residual degradation can become audible. The 20-bit solution seems to strike an optimal balance of perfectionist sound, with some 10 dB of safelty margin for downstream processing, without excessive waste in the payload.