Due to recent conversations in other threads I decided to refresh my knowledge of high frequency hearing thresholds. Its been a long time since I looked at it, so it was a good time to see what newer research has found. The attached document is one of the ones I looked at. Its a hearing threshold study performed by the Ear Research Group, Department of Otorhinolaryngology, Hospital Universitario Puerta de Hierro, Universidad Aut ó noma de Madrid, Madrid Spain, of 645 subjects.
https://en.wikipedia.org/wiki/Autonomous_University_of_Madrid
It appears to be endorsed by the British, International and Nordic Audiology Societies . It also compares its results to those of multiple other referenced previous studies. Threshold results in this study were generally lower than previous studies. I will let you peruse the study at your leisure.
First point, obviously 20kHz isnt a brick wall limit to hearing, I never meant to imply it was in the other discussion.
Lets look at the data. What we see on page 7 is that the best results, the 5 to 19 year age group, had a hearing threshold of 65dB at 20kHz. Previous studies showed higher levels of 77 to 105dB (page 11). In simple terms this means that to detect a single tone at 20kHz sound levels need to be at least 65dB. Subjects of 20 to 29 years old its 85dB, 30 to 39 years its 91dB, 40 to 49 years its 97dB and 50 to 59 years its 104 dB. (thats a depressing trend for us older forum members!)
You might react by saying "oh 65dB isnt that loud". Well yes, granted, however the level of musical content at 20kHz is vastly lower than at lower frequencies.
I have subsequently examined a selection of musical tracks in Adobe Audition (more than in the other thread). I high pass filtered 20 to 22 kHz and used the amplitude statistic function to see the levels. This is time domain as to avoid issues with FFT effectively time averaging and reducing values. Highest level I found was with Beuna Vista Social Club track at - 31dB. Lots of brass and percussion so not unreasonable to expect lots of high frequency content. Note that this is the highest peak signal level, it is NOT an indication of the level of any individual tone. Individual discriminate tones will be at a lower level.
If you look at the plot near the marker you will see a tiny area that is louder than in rest of the track.
So if the track was being played at a general listening level of about 95dB, which is fairly loud, the best performing group might be expected to hear something in that 20 to 22kHz frequency range. Thats 95dB minus the 30dB lower HF content to = 65dB hearing threshold. However, this ignores the fact that the sounds would be masked to an extent by louder other sounds elsewhere in the frequency range. So this is is a best figure, not a realistic figure for actual audibility of 20 kHz sounds. Also, remember the peak figure is not for any discriminate tone. The Adobe measured "perceived" loudness FWIW was around -53dB, significantly lower again. These are also the lowest threshold figures I found. The figures are, as mentioned, higher for other studies referenced in this one.
So going back to the question of the audibility of transition band aliasing. This is higher frequency again, above 22kHz. Hearing sensitivity has fallen again for this region. I dont have figures for that but would guestimate a good 5dB, maybe more.
If there is any reduction in signal level within the band, and member @Blumlein 88 measured up to -85dB towards 20kHz IIRC, IMO it starts to become very, very difficult to accept the notion that it is in any way audible - even to the people that are at the "right" end of the bell curve. Any of us listen beyond 110dB? No, and if you do you certainly wont be hearing 20kHz for much longer.
Any age group older than 20 years has absolutely no chance whatsoever.
OK, what about transition band intermodulation appearing in the normal audio band. Well below is an NC400 amplifier IM plot. 19.5 and 18.5 kHz tones. IMD products are at least 100dB down.
So 100dB down on signals that are already low and more like -130dB down in the sensitive part of the hearing range.. Again its a non issue.
https://en.wikipedia.org/wiki/Autonomous_University_of_Madrid
It appears to be endorsed by the British, International and Nordic Audiology Societies . It also compares its results to those of multiple other referenced previous studies. Threshold results in this study were generally lower than previous studies. I will let you peruse the study at your leisure.
First point, obviously 20kHz isnt a brick wall limit to hearing, I never meant to imply it was in the other discussion.
Lets look at the data. What we see on page 7 is that the best results, the 5 to 19 year age group, had a hearing threshold of 65dB at 20kHz. Previous studies showed higher levels of 77 to 105dB (page 11). In simple terms this means that to detect a single tone at 20kHz sound levels need to be at least 65dB. Subjects of 20 to 29 years old its 85dB, 30 to 39 years its 91dB, 40 to 49 years its 97dB and 50 to 59 years its 104 dB. (thats a depressing trend for us older forum members!)
You might react by saying "oh 65dB isnt that loud". Well yes, granted, however the level of musical content at 20kHz is vastly lower than at lower frequencies.
I have subsequently examined a selection of musical tracks in Adobe Audition (more than in the other thread). I high pass filtered 20 to 22 kHz and used the amplitude statistic function to see the levels. This is time domain as to avoid issues with FFT effectively time averaging and reducing values. Highest level I found was with Beuna Vista Social Club track at - 31dB. Lots of brass and percussion so not unreasonable to expect lots of high frequency content. Note that this is the highest peak signal level, it is NOT an indication of the level of any individual tone. Individual discriminate tones will be at a lower level.
If you look at the plot near the marker you will see a tiny area that is louder than in rest of the track.
So if the track was being played at a general listening level of about 95dB, which is fairly loud, the best performing group might be expected to hear something in that 20 to 22kHz frequency range. Thats 95dB minus the 30dB lower HF content to = 65dB hearing threshold. However, this ignores the fact that the sounds would be masked to an extent by louder other sounds elsewhere in the frequency range. So this is is a best figure, not a realistic figure for actual audibility of 20 kHz sounds. Also, remember the peak figure is not for any discriminate tone. The Adobe measured "perceived" loudness FWIW was around -53dB, significantly lower again. These are also the lowest threshold figures I found. The figures are, as mentioned, higher for other studies referenced in this one.
So going back to the question of the audibility of transition band aliasing. This is higher frequency again, above 22kHz. Hearing sensitivity has fallen again for this region. I dont have figures for that but would guestimate a good 5dB, maybe more.
If there is any reduction in signal level within the band, and member @Blumlein 88 measured up to -85dB towards 20kHz IIRC, IMO it starts to become very, very difficult to accept the notion that it is in any way audible - even to the people that are at the "right" end of the bell curve. Any of us listen beyond 110dB? No, and if you do you certainly wont be hearing 20kHz for much longer.
Any age group older than 20 years has absolutely no chance whatsoever.
OK, what about transition band intermodulation appearing in the normal audio band. Well below is an NC400 amplifier IM plot. 19.5 and 18.5 kHz tones. IMD products are at least 100dB down.
So 100dB down on signals that are already low and more like -130dB down in the sensitive part of the hearing range.. Again its a non issue.
Attachments
Last edited: