View Full Version : Digital Facts and realities?
To add to my comment about higher sampling rates.
Our hearing has a sensitivity to sound measured at 5-10 us(micro seconds) but a sampling rate of 44.1 khz only gives us 22.7 us.
96 khz sampling rate gives a time resolution of 10.4 us but is not until you get to 24/192 khz sampling rate that you get a resolution of 5.2 us which is close to the time resolution of our hearing.
Therefore CD quality sampling rate may not be enough and may be one reason our hearing can perceive timing issues.
This is not to be confused with timing of the digital audio signal through poor clocking, RFI or a poor PSU resulting in jitter and corruption of the audio data.
None of this may be of any concern however to those of us who have poor or knackered hearing!:)
To add to my comment about higher sampling rates.
Our hearing has a sensitivity to sound measured at 5-10 us(micro seconds) but a sampling rate of 44.1 khz only gives us 22.7 us.
96 khz sampling rate gives a time resolution of 10.4 us but is not until you get to 24/192 khz sampling rate that you get a resolution of 5.2 us which is close to the time resolution of our hearing.
Therefore CD quality sampling rate may not be enough and may be one reason our hearing can perceive timing issues.
This is not to be confused with timing of the digital audio signal through poor clocking, RFI or a poor PSU resulting in jitter and corruption of the audio data.
None of this may be of any concern however to those of us who have poor or knackered hearing!:)
Thank you! Exactly what I’ve been saying, there is greater accuracy to be had well within human hearing. Higher frequencies is not the only advantage of higher rates. Those with knackered hearing may not appreciate higher frequencies, but way down in the midrange where most of the music is, it can make a real world difference.
Russell
Yes thanks James, this is what I've been trying to get to over the last 5 pages of posts.
To add to my comment about higher sampling rates.
Our hearing has a sensitivity to sound measured at 5-10 us(micro seconds) but a sampling rate of 44.1 khz only gives us 22.7 us.
96 khz sampling rate gives a time resolution of 10.4 us but is not until you get to 24/192 khz sampling rate that you get a resolution of 5.2 us which is close to the time resolution of our hearing.
Therefore CD quality sampling rate may not be enough and may be one reason our hearing can perceive timing issues.
This is not to be confused with timing of the digital audio signal through poor clocking, RFI or a poor PSU resulting in jitter and corruption of the audio data.
None of this may be of any concern however to those of us who have poor or knackered hearing!:)
Where do you get that figure? A rise time of 10us (i.e. a half period) implies 50kHz. 10us might be enough for the ear to respond to an aural stimulation, but to recognise the pitch of a note, the ear requires a longer time; at least 5 cycles.
https://pdfs.semanticscholar.org/381e/63313255d53c2ddabb973128dd6bd04d5112.pdf
The figure of 10us probably relates to the perception of inter-aural time differences, which enable us to locate sounds. Obviously this is important because if the sound system can't resolve at that level, the auditory system will be confused.
Where do you get that figure? A rise time of 10us (i.e. a half period) implies 50kHz. 10us might be enough for the ear to respond to an aural stimulation, but to recognise the pitch of a note, the ear requires a longer time; at least 5 cycles.
https://pdfs.semanticscholar.org/381e/63313255d53c2ddabb973128dd6bd04d5112.pdf
I was reading a number of articles Barry but I think it was buried somewhere in this one.
https://asia-latinamerica-mea.yamaha.com/en/products/contents/proaudio/docs/audio_quality/04_audio_quality.html
That's an interesting paper, thanks. However if we take the speed of sound to be 330m/s, a time resolution of 6um would imply the subjects of Kunchur's experiment were able to tell if one loudspeaker had been moved 2mm towards the listener relative to the other. I find that hard to believe.
Light Dependant Resistor
01-04-2019, 01:39
An interesting moment to reflect on comments left by engineers and others here: http://www.aes.org/historical/oral
That's an interesting paper, thanks. However if we take the speed of sound to be 330m/s, a time resolution of 6um would imply the subjects of Kunchur's experiment were able to tell if one loudspeaker had been moved 2mm towards the listener relative to the other. I find that hard to believe.
I can hear the difference when I move my speakers by that much.:lol:
Light Dependant Resistor
01-04-2019, 05:58
I can hear the difference when I move my speakers by that much.:lol:
But depends also on the loudspeaker used and size of the room and furnishings . For instance a ESL57 you may be able to pick up a 2mm difference in a small room with minimal
or no furnishings other than a chair , possibly noting more difference than cone speakers that were moved the same distance.
And of course it presupposes you listen with your head held in a clamp, so it can't move. ;)
Light Dependant Resistor
01-04-2019, 23:04
And of course it presupposes you listen with your head held in a clamp, so it can't move. ;)
I think a head clamp might add a unnecessary boundary reflection, possibly obscuring or confusing the anticipated 2mm difference.
Maybe try each, to see. :)
Even if you multiply by ten, it’s still a very small amount. If one can hear a speaker moved 20mm that’s still significant.
Russell
Primalsea
02-04-2019, 07:54
Two interesting articles on Sound on Sound:
https://www.soundonsound.com/techniques/digital-myth
https://www.soundonsound.com/techniques/mqa-time-domain-accuracy-digital-audio-quality
Generally information on the SOS website can be considered reliable, although some might argue that the second is some disguised marketing.
One thing that I did find of particular interest was the idea that although our brains can process timing information very well to the point where we may not be able to perceive these “timing errors” if our brains are required to work less in this regard then quite possibly extended or critical listening may well be more satisfying.
Even if you multiply by ten, it’s still a very small amount. If one can hear a speaker moved 20mm that’s still significant.
Russell
I find it depends on the speakers. My OBs much more sensitive to fore and aft placement than box speakers, presumably due to the sound emitting in both directions so there's a more complex relationship between direct and reflected sound. I do indeed fine tune the speakers to a very small number of cms, could well be around 2 cms, but not 2mm!
More generally yes 2mm is going too far in reality. On the one side we have sampling at 22.7uS being thought of as easily good enough. On the other side we have suggestions that 5 to 10uS may only just be enough to allow us to be relaxed and sense the best timing.
Has either suggestion been fully put to scientific trials? And no I don't mean DBT because if we are talking about relaxed listening and sensing timing I don't believe that typical DBT tests are any use whatsoever. We'd need a clever scientist to come us with a good test or do we end up needing to rely on a large number of subjective tests - long term listening in a controlled environment?
Two interesting articles on Sound on Sound:
https://www.soundonsound.com/techniques/digital-myth
https://www.soundonsound.com/techniques/mqa-time-domain-accuracy-digital-audio-quality
Generally information on the SOS website can be considered reliable, although some might argue that the second is some disguised marketing.
One thing that I did find of particular interest was the idea that although our brains can process timing information very well to the point where we may not be able to perceive these “timing errors” if our brains are required to work less in this regard then quite possibly extended or critical listening may well be more satisfying.
Interesting point. What we know about dyslexics gives us some clues. Many dyslexics suffer overload in terms of information processing of the auditory. The brain cannot process sounds fast enough so words may not be fully broken down eg marmite could be heard as mumit, often it's the middle of the work that's an issue but it does vary. Bear mind that about 10% of us exhibit some dyslexia. Maybe dyslexics need a higher sampling rate to allow them to register timing well - but maybe that's presents more information making things worse....so many questions!
Primalsea
02-04-2019, 08:52
If only audio engineers and psychologists got together!
The blind usually have sharper hearing so may be the best subjects.
Up and down a few mm will make a difference. Doubt if it would back to front distance tho
The blind usually have sharper hearing so may be the best subjects.
Up and down a few mm will make a difference. Doubt if it would back to front distance tho
I feel we need a spectrum of people. What's best for one is unlikely to be best for another. I'm sure we've all experienced people having very different musical presentations preferences to own preference. It's a bit like dieting, not all metabolisms are the same. People with excellent hearing and the best phonological processing - ability to distinguish the sounds efficiently - will I suspect have differing requirements to the dyslexics I've described. Maybe there isn't a gold standard perfect system, just one that suits specific individuals and we could do with understanding what characteristics suit certain type of people.
I feel we need a spectrum of people. What's best for one is unlikely to be best for another. I'm sure we've all experienced people having very different musical presentations preferences to own preference. It's a bit like dieting, not all metabolisms are the same. People with excellent hearing and the best phonological processing - ability to distinguish the sounds efficiently - will I suspect have differing requirements to the dyslexics I've described. Maybe there isn't a gold standard perfect system, just one that suits specific individuals and we could do with understanding what characteristics suit certain type of people.
Interesting point. I feel that most double blind tests are of no use, because the test group is of an average audience. The test subjects range from the super golden ears, to the next to deaf, and all in between. Why would we want to test high end stereo using near deaf testers? Hearing tests should be performed on all those considered for the test, and only those of the highest results used in the test. Otherwise, we are always going to come to the same conclusion. That results are no better than chance.
. We aren’t testing to see if the average person can hear it, we are testing to tell if there is a difference! I see it time and again that these lax tests are said to prove that there is no audible difference, and perhaps marketers of mass production equipment are only interested in what the average person hears? But those who design high end gear are not concerned about what average people hear, they are concerned with what the few who are willing to pay top dollar for sound quality can hear.
Russell
Thing is, you don’t need a spectrum of people to achieve statistical significance. If just one person can spot the difference significantly, then there is a difference. Scale the test to another 5 people, who all fail it, human nature...
Thus, I think the best type of ABX test is at home, in your own system, your own music, not some strange lab.
End of the day, most people on this forum are very experienced listeners, who look out for things that most don’t, to a fault sometimes he he.
Powered by vBulletin® Version 4.2.3 Copyright © 2026 vBulletin Solutions, Inc. All rights reserved.