Unfortunately I can't remember exactly where I read the 16/24 bit experiment, which is a shame because it had some interesting stuff about how it depends heavily on the reproduction system and that not everyone can tell the difference (honest!).SY said:
I've been searching Google to see what I can come up with, but it doesn't seem to be there (it was in a book I read, not on the internet). The best I could find was this page, which implies that double-blind tests have been done to show that there is a difference between 16 and 20 bit, but not 20 and 24bit, and also another page which describes an ABX test showing audible differences between 16 and 24bit mp3s. I have no idea how reliable that second one is.
Thanks, you're apparently not so evil! These things tend to be slippery and audio Urban Legends take on a life of their own.
Now, there is a circumstance where I think it would absolutely be reasonable that the difference between 16 and 24 bits is detectable by ear- when the input signal amplitude is deliberately limited so that the top several MSBs are not used. Admittedly, it's a bit of a phony situation, but I've certainly seen worse.
16 bits gives a dynamic range of better than 90 dB, so I'm, ahem, skeptical of claims of audibility differences in a non-lab environment. I'd be surprised if in a lab setting, there could be a difference to the ear, but I think that's certainly within the realm of possibility.
Now, there is a circumstance where I think it would absolutely be reasonable that the difference between 16 and 24 bits is detectable by ear- when the input signal amplitude is deliberately limited so that the top several MSBs are not used. Admittedly, it's a bit of a phony situation, but I've certainly seen worse.
16 bits gives a dynamic range of better than 90 dB, so I'm, ahem, skeptical of claims of audibility differences in a non-lab environment. I'd be surprised if in a lab setting, there could be a difference to the ear, but I think that's certainly within the realm of possibility.
gmarsh said:
The most uncontrollable part is during live recording. Once you get compresson or limiting, there is no way to recreate the clipped part. I remember building a "Peak unlimiter" many years ago and tested it on piano music, it did make the piano sound more dynamic.
Is DVD Audio really better? I remember sales mentioning different recording methods for 7.1 channels that would protentially recreate the actual sound feild in a concert hall.
real said:
Actually the human hear is band pass.
If you just try to listen to steady frequency signals, the upper limit might be close to 20K depending on the individual. If you listen to transients that can be decomposed to indicate 50K content, many humans can detect the difference be tween exsitence of the 50K and lack of.
I remember having an ultrasonic mosquito repelling device, while I could hear the sound it produced about 5 inches away, some people could not hear it even if you stuck it in their ear.
You are not very specific about the transient - is 50k the lowest frequency component? Many transients (eg pulses) have lower frequency content in them too. If 50k is the lowest frequency, could you provide an example of the transient, eg its shape and how it might appear in music?
Regards,
Chris.
Regards,
Chris.
Euphase said:
Interesting paper. I wonder how it works in our hearing and heads.
Maybe the high frequencies do something to "pre-stimulate" the hair-cells in the vestibulae.
It explains (to my opinion) however why good SACD's indeed sound "softer".
Don't worry about the pitfalls in the simple test, its not academic.
It is however interesting, easy to check on yourself and your neighbour and at the least gives you a stimulus to think "why the heck is this happening".
marconist said:
Don't worry about the pitfalls in the simple test, its not academic.
It is however interesting, easy to check on yourself and your neighbour and at the least gives you a stimulus to think "why the heck is this happening".
May be you should really think why the heck it is happening? I have given you the reasons why the heck it might be happening, very valid reasons. I tried to be polite so far, but you really have not a full understanding of the signals, sampling, reconstruction etc, and yet having a position as if you know it all and dispersing advice, calling others valid points as "noise" etc.
Just a simple thing to think about: Why don't you ask to yourself, if the test to validate that humans on average hear things about 20Khz is so easily verifiable with the test you mentioned, why all these people in the research field go through the trouble of even measuring brain activity to figure out whether humans really respond to spectrum above 20Khz or not? May be you will tell them the same thing, do that simple experiment you mentioned and think "why the heck is this happening?". Like they didn't perform it anyway... Show me a rigorous, peer reviewed and widely accepted research results that show that humans HEAR above 20Khz. This paper is close, but it doesn't show that we hear above 20Khz, it just shows (if there weren't any other things to cause it) our brains respond to sounds above 20Khz, even if we don't know it. I would like to know.
Let me see, where we started and what are we know. First the claims were that because there were only 2 sampling points per 20Khz, it couldn't reproduce signals containing 20Khz components. Then it turned into, it is inferior because humans can detect beyond 20Khz and 40Khz sampling requires filtering above 20Khz, therefor it causes audiable data to be lost. This thread is degenerating but running in circles while doing so......
soongsc said:
Who, where, when, with what equipment, under what circumstances? Or is this just another anectodal we golder ears can hear it? Really, I would like to know if this has been already verified with many humans. But I don't want to hear about uncontrolled tests, or tests done with nonlinear equipment which alters the frequency spectrum or poorly designed tests that alter the 10-20Khz range even in a linear way.
Here is ainteresting reading where Ivor Tiefenbrun (aka Linn) faced some of his claims in a blind test:
http://www.bostonaudiosociety.org/bas_speaker/abx_testing2.htm
To summarize ther digital test briefly:
"Tiefenbrun has been quite outspoken about the inadequacies of digital audio recordings and the systems on which they are made, and his organization, Linn Products Ltd., was instrumental in publishing an analysis of the Sony consumer PCM-F1 digital audio adapter outlining their objections...
...John Vanderkooy have moreover conducted blind listening comparisons between the PCM-F1's ( a first generation PCM encoder/decoder) input and its reconstructed output signals, and had yet to find anyone (including Ivor Tiefenbrun ) who can reliably distinguish between them on musical program material."
http://www.bostonaudiosociety.org/bas_speaker/abx_testing2.htm
To summarize ther digital test briefly:
"Tiefenbrun has been quite outspoken about the inadequacies of digital audio recordings and the systems on which they are made, and his organization, Linn Products Ltd., was instrumental in publishing an analysis of the Sony consumer PCM-F1 digital audio adapter outlining their objections...
...John Vanderkooy have moreover conducted blind listening comparisons between the PCM-F1's ( a first generation PCM encoder/decoder) input and its reconstructed output signals, and had yet to find anyone (including Ivor Tiefenbrun ) who can reliably distinguish between them on musical program material."
So,
non-os without filtering ?
non-os with steep filtering to get rid of the distortions caused by the staircases ?
8 x os with interpolation to make a low order filter possible ?
I read a few times that steep filters sound bad especially when containing coils, others say they are expensive, what is true ?
non-os without filtering ?
non-os with steep filtering to get rid of the distortions caused by the staircases ?
8 x os with interpolation to make a low order filter possible ?
I read a few times that steep filters sound bad especially when containing coils, others say they are expensive, what is true ?
Both and neither are true, depending on where and how they are used. Blanket statements like this are pretty commonly thrown around.
Now, in the context of the question at hand, i.e., satisfying the Nyquist criteria so that the system functions properly, you're really talking about the steepest practical filter as close to the Nyquist frequency as you can get. Putting aside the wonderful tricks possible in the digital domain to make reconstruction easier, if you believe that phase shift of the sort caused by a steep lowpass filter at (say) 21kHz is audible, it is trivially easy to compensate for.
real said:
Currently information that I have shows 14" cymbals can go beyond 20K, How high it gets, I don't know. I suppose if it mattered to anyone, they could just do their own test to verify.
As far as the 50K thing, I would find out where it came from if it would make a difference to the audio world than just winning a debate.
Euphase said:
I think it would make no difference whether the information was presented or not. Since it would be necessary to go through material more than 20 years old, and a pile of data. If someone has recently conducted tests to prove the contrary, it would be interesting to hear about. Then we can figure out whether its a problem with the tests, or if this generation is gradually suffering hearing lose in some way.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.