I don't believe that training can allow you to hear frequencies above 20kHz nor noise floors below -90dB.
I tend to agree that this is probably the case in normal listening conditions. It would be interesting to see if there has been any testing on this though.
"Training" is just a way to try to get people to not trust their own ears and take your word for it. It's a power play that has nothing to do with perceptual ability.
Perhaps not surprisingly, I would disagree with this. At least when it comes to detecting things like audible noise and distortion, and some types of compression, resampling or aliasing artifacts. I can't give you any demonstrable proof of this though, beyond my own opinion. So my assertion is basically meaningless from a scientific standpoint.
I try not to comment on this guy, but when he made the claim to me that he could hear the difference between 16 and 24 bit, I called him on goosing the volume and he refused to admit to me that he was doing gain riding. Cranking the volume on quiet parts isn't some unique skill that takes training. Sure you can hear something down below -90dB if you pump up the volume way up on a fade out. That doesn't take training. That is just plain cheating. No one listens to music riding the volume level to pull up all the quiet parts. Parlor tricks.
Fwiw, he acknowledges much of this in the above video. And it's why I gave it the label "extreme instances". Because this is certainly not the way most people listen to music.
If he were arguing the point with you though, then I think he might point out that "totally transparent in
all listening conditions" and "transparent under normal listening conditions" may be two different things. And if you want the former rather than the latter, then maybe some of these smaller details may be worth paying attention to. It's just bits after all, so it's not like a few more is gonna hurt anything.
And maybe, in some rare or extreme circumstances, it might actually help a little.
If Amir can pick out the differences in quieter recordings at a higher volume, then maybe... just maybe, it's also a possibility for others to detect some noise or quantization errors in some lower volume 16-bit recordings as well. (?)
I believe the person who could hear the highest was a child who could hear up to 23kHz. That isn't even one whole note on the musical scale above 20kHz.
...Which is why I mentioned that sound in that range is mostly overtones and timbral in nature. I certainly can't hear that high though. It would be interesting to see if there have been more studies on this as well though. Perhaps there are certain demographic groups with slightly wider or higher ranges of hearing than others, for example.
Some also claim that they can "feel" the higher frequencies as well. Which is certainly the case with frequencies below 20 Hz. I dunno about the higher frequencies though.
All this stuff about super audible frequencies is bologna. 16/44.1 is already overkill. The only reason higher data rates are needed is for mixing and mastering. None of the exceptions you mention have any relevance to listening to music in the home. The difference is inaudible in the real world outside the heads of audio "experts".
I disagree somewhat with you on this as well. Mainly because of the proliferation of higher-quality tools, digital tools that is, for manipulating and altering audio content more to preference on the user's end. In days of yore, this kind of thing was done mostly in the analog domain.
If higher bit depths and sample rates are beneficial for mixing and mastering content. Then why not also for making similar kinds of adjustments in the digital domain on the user's end? That is my only point (I think).
Video chose 48kHz because it divided evenly with the frame rate of film. It made film to video conversions less difficult.
Yes! Now that you mention it. I seem to recall reading that somewhere as well. Thank you for the correction on that.
