After watching the video interview with Scott, Mark seems to put more emphasis on bit depth and dynamic range and I agree with everything on this topic. Also the improper and often ineffective implementation of HD audio by many manufacturers/services that claim to support it, and the importance of at least providing the fidelity you are advertising, again I agree here. What was said about the Pono music service was particularly informative and interesting but not at all surprising, thank you for that Mark.
We agree that a higher sampling rate makes sense throughout the mastering process for various reasons. We agree (I think) on the implications of the sampling theorem. What you said at around 55:00 into the video about differences in hearing I agree with, but the exception is hearing or perceiving anything above about 20KHz which is impossible for humans. The Nyquist Shannon sampling theorem isn't a property or artifact of our particular digital encoding process, it wasn't intended to facilitate digital audio recordings, its a fundamental axiom of universal information theory. Which means it also applies to the brain's encoding of audio back to electrical signals (whatever they may be, irrelevant that they're not digital), so encoding that information is indeed a waste of bits. Which itself, I admit, is not a terribly relevant point since we are not constrained by storage capacity anymore, if everyone agreed on 96/24 that would be just swell, just aslong as we don't start a sampling rate or bit depth war (as an extension of 'loudness wars') amongst these profit-minded companies that inflates files to ridiculous sizes.
: The part I wish to question is where you suggested that people could hear/perceive sounds above 20KHz, (at 12:45 in the video interview). Could you elaborate? What is your evidence for this?
In general, I do not trust my ears or what anyone else's ears tell them, hearing is loaded with cognitive biases, controlling for all of them with proper testing is neigh on impossible, singling out the minority of studies that have done it correctly is not a good use of one's time. For example I cannot tell the difference between 320Kbps .mp3 vs CD vs 96/24 flac (for most
audio tracks). However I can logically reason with my knowledge of maths and audio theory and human hearing that there is difference between mp3 and CD that some people should be able to hear even if I can't. Similarly I can reason that there is a difference between 48KHz sampling rate and 96KHz, the latter encodes twice the bandwith, but I cannot
reason that any human
should be able to hear it. If the error is with my knowledge or reasoning, I am listening, please enlighten me (anyone). I am loathe to trust any evidence that takes the form of listening tests asking subjects to say what sounds 'better', no matter how well controlled, no matter how good the methodology looks on paper, hearing is loaded with bias, the testing is loaded with bias given the profit margins acheivable in the high-end audio industry. A listening test that asks 'do these sound different?' is marginally better.