Quote:
I wish I could speak French half as good as you can speak English. There are actually numerous studies and this threshold has been reported to be anywhere from 5-10µs. Here is a link to one of those studies and it references others:
http://boson.physics.sc.edu/~kunchur/temporal.pdf
Here is also a video by Hans Beekhuyzen that you might find helpful:
https://www.youtube.com/watch?v=r_wxRGiBoJg
thanks. the video wasn't relevant to me(ok vulgarisation though, I was expecting more BS from someone pushing for his product). but the paper was nice. at last I know how those crazy low numbers are obtained, and it has nothing to do with listening to music.
they use iterated ripple noise in one case, and square waves in the other, none of which you can really hope to find in natural sounds. still the results seem to be relevant in the context of the experiment. it's trying to use them as is, in the context of listening to music, that doesn't work with me at all.
with the variety of sounds in music, all the possible masking, and the little musical content in high frequencies, I'm not sure there actually is a need for such sampling speed for most music content. but more than that, I highly doubt the average human could hope to still react to such low temporal thresholds while listening to music.
another example that come to my mind to explain why I believe this: maximum audible dynamic range.
when testing it, we can rapidly demonstrate in an anechoic chamber that the average human can hear very very low sounds at, let's say 1db to make it simple. it is then very easy to push the loudness until it hurts, somewhere in the 130 or 140db. and there we have it, most humans can hear up to almost 140db of dynamic!!! that's true within that given test. not at all true when listening to music.
with average musical content, sounds of different loudness are coming in at once, not one, then silence, then the other. in this case most humans will struggle to discriminate more than the first 60db of music in instantaneous dynamic. and that's a figure that actually relates to my own experience of music. if I add some sounds 80db below most tracks, I fail to notice the added sound almost all the time. so we can't just take the 140db figure and slap it onto what music needs to do to sound right. it's a misuse of data.
I'm confident it's the same with timing, first because again there are no convenient cues like IRN or square waves in a song, so it is obvious the threshold will crumble with real music.
but also because the music will most likely be complex with many temporal cues of many different nature. when the brain is working on getting the pitch, the position, recognizing the song, the instruments... we fall in the drop of performance due to multitasking and the question of the brain focusing on what matters most to it and rejecting most of the information coming into the ear.
and that aligns with the typical failure of controlled tests to discriminate 16/44 vs highres. while your numbers and logic do not. if all the stuff between 7µs ans 20µs was really noticed with real music content, why would we fail so easily to notice them in a blind test?
I see this as just another out of context misuse of a valid science experiment. all this does not make a case against highres, but it also fails IMO to make a valid case against CD quality not being enough.