I certainly can't parse the "bandwidth" statement, in digital audio you can only represent frequencies up to 1/2 the sample rate of the .wav file
in fact this thread's whole recent digital audio, digitized pulses, hearing wav file differences has me confused as to the point re John Curl's persistent claims that digital audio isn't "good enough" to interest him/show the "subtle phase errors" he keeps proposing that are "immeasurable" but responsible for the superior sound he can get when (lack of) price or objective performance constraints allow him a free hand with "low feedback", discrete designs
can't we take it as given that human hearing as understood by conventional psychoacoustics has a "zoo" of "odd" resolution capabilities - Amazon.com: Psychoacoustics: Facts and Models (Springer Series in Information Sciences) (9783540231592): Hugo Fastl, Eberhard Zwicker: Books comes with a demo CD
but doesn't that strongly suggest that these however curious psychoacoustic effects "fit" in "conventional" audio media, dynamic range an bandwidth? are apparently reproduced well enough to illustrate the book’s points on at best indifferent quality equipment by audiophile standards?
If I were trying to “justify” why 10% THD @ full output SET amps don’t necessarily “sound bad” the wealth of perceptual masking phenomena is really useful
But I don’t see the logic/relevance of demoing phase variations of 30 degrees at 9 KHz to the recurrent question of why well designed high accuracy, high feedback amps apparently “do something wrong” – with emphasis in John Curl’s arguments that is a some immeasurable error on the part of the high feedback amps with excellent "known" measurements (IMD, PIM, slew rate, multitone IMD, noise fill... Hirata? ) and not just a preference for added distortion that makes “simple” low feedback/no global feedback designs audibly superior
in fact this thread's whole recent digital audio, digitized pulses, hearing wav file differences has me confused as to the point re John Curl's persistent claims that digital audio isn't "good enough" to interest him/show the "subtle phase errors" he keeps proposing that are "immeasurable" but responsible for the superior sound he can get when (lack of) price or objective performance constraints allow him a free hand with "low feedback", discrete designs
can't we take it as given that human hearing as understood by conventional psychoacoustics has a "zoo" of "odd" resolution capabilities - Amazon.com: Psychoacoustics: Facts and Models (Springer Series in Information Sciences) (9783540231592): Hugo Fastl, Eberhard Zwicker: Books comes with a demo CD
but doesn't that strongly suggest that these however curious psychoacoustic effects "fit" in "conventional" audio media, dynamic range an bandwidth? are apparently reproduced well enough to illustrate the book’s points on at best indifferent quality equipment by audiophile standards?
If I were trying to “justify” why 10% THD @ full output SET amps don’t necessarily “sound bad” the wealth of perceptual masking phenomena is really useful
But I don’t see the logic/relevance of demoing phase variations of 30 degrees at 9 KHz to the recurrent question of why well designed high accuracy, high feedback amps apparently “do something wrong” – with emphasis in John Curl’s arguments that is a some immeasurable error on the part of the high feedback amps with excellent "known" measurements (IMD, PIM, slew rate, multitone IMD, noise fill... Hirata? ) and not just a preference for added distortion that makes “simple” low feedback/no global feedback designs audibly superior
Last edited:
Most audio power amplifiers have a dominant pole in their response that limits the high frequency response. So for those people who can hear 15 degrees of phase shift at 9 kilohertz, what single pole bandwidth is required? Now what bandwidth is required for 5 degrees at 20 k?
Note the .wav file is played through a multipole LPF so that the bandwidth calculation is not the same.
Note the .wav file is played through a multipole LPF so that the bandwidth calculation is not the same.
When generating digital content algorithmically, which I assume was done to create this phase-modulated sine wave, there is a very real risk of creating aliased frequencies. While it is true that digital audio can only represent frequencies up to Nyquist, the trouble is that nothing automatically removes the higher frequencies from generated waveforms. If you create a pure square wave in a .wav file, it will have infinite frequencies (to the limit of the bit depth's ability to resolve the lowest levels), but any of those frequencies above Nyquist will simply be aliased down between Nyquist and DC - in other words, you won't hear a pure square wave on output. The reconstruction filter after the DAC will not be able to remove these frequencies because they will have already been aliased in the data itself.
In other words, my very first assumption would be to doubt the data, at least unless its provenance can be verified and duplicated.
There is most certainly a preference for added distortion which ties directly to the known preference for louder but otherwise identical sources. The human hearing system perceives loudness differently than a VU meter, peak-reading meter, or any (simple) electronic circuit. There are critical bands which affect our perception, such that added tones within an existing band contribute less to our perception of loudness than tones added in previously-empty bands. The interesting thing about harmonic distortion is that it has great potential to add tones in higher critical bands that might have been silent, and thus our perception of the loudness is that the distortion has increased the level.
Given the known factor that volume must be precisely matched between components before their true perceived quality can be fairly determined, and the fact that (nearly all present-day) meters do not represent loudness exactly the same as we perceive it, then it makes sense that there would be false preferences for distortion which are effectively nothing more than the old preference for louder sources.
I'm out @16kHz. Famous 17.4 kHz Mosquito sound, I can only "sense" that sound, not "hear".
Both my children can hear that sound without problem and they are screaming if I play that sound too loud
Another interesting LINK
Both my children can hear that sound without problem and they are screaming if I play that sound too loud
Another interesting LINK
Last edited:
What amazes me is WHAT added distortion that I might add to my designs like the JC-1, or the JC-2, two well documented designs, independently measured, and how much I would like?
NO, I work toward LOWER amounts of negative feedback, because the results sound more 'natural' to me. Gets me pretty good opinions from the reviewers, also.
I don't WANT to add distortion, BUT I do not believe the .01% of low order harmonic is audible, as it would be masked by the much greater distortion of the loudspeakers or the sources. Since, I can get 0.01% distortion at realistic output levels, I am happy with as little feedback as possible, sometimes even 'open loop'.
NO, I work toward LOWER amounts of negative feedback, because the results sound more 'natural' to me. Gets me pretty good opinions from the reviewers, also.
I don't WANT to add distortion, BUT I do not believe the .01% of low order harmonic is audible, as it would be masked by the much greater distortion of the loudspeakers or the sources. Since, I can get 0.01% distortion at realistic output levels, I am happy with as little feedback as possible, sometimes even 'open loop'.
Last edited:
A sudden change in phase as in one sample period to the next violates the sampling theorem in the same way as a sudden amplitude change. Try to think about going from 0 to 5 degrees of phase at 20kHz in a way that there are NO frequency components beyond 22.1kHz. It's a slow dribble.
Scott,
It is good to see we disagree on everything. Last time we came close to agreeing the world almost ended.
What Schroeder and Neve have mentioned is that 5 degrees of phase shift at 20K in an amplifier sounds different than no phase shift caused by the amplifier.
Schroeder in particular did his test by generating DSP tones making sure the peak amplitude did not change.
The .wav file I posted had the same frequencies in the same ratio just with different phase shifts. If you want to worry about how the splice point affects things go right ahead, it is of no importance to the actual issue.
I assume you understand the idea of a dominant pole and how that affects phase shift. I also assume the calculation of what single pole roll off bandwidth is required to keep the phase shift at 20K below 5 degrees is not very hard for most folks participating in this thread.
SY,
And how would you do a double blind Alien Abduction test?
If folks tell me they can't hear any change that is fine. If they tell me they have never been abducted by aliens that is fine. If they tell me they can hear the increase in harshness, that corresponds to what we hear here. If they tell me they have been abducted by aliens, I want details on their technology!
ES
Yes I have tried to see what phase shift I can hear. I do not get to 5 degrees at 20K, but I do seem to get 15 at 10K.
Have you tried to ABX crystal energy and frozen photographs? You could try it, or ignore it, or make fun of it because it seems so ridiculous. But every so often ridiculous turns out right. (OK very very rarely.)
Yes it is simple to test phase change perception, but who do you know who has actually tried it?
There's quite a rich literature on it.
Yes, right after I try sending my bank account info to that nice Nigerian gentleman.
I would try w/wo 5 degrees at 20k (EDIT - BOTH channels) while listening to familiar music. I have tried w/wo 17kHz brick wall HUGE phase shifts and heard nothing. Dave Greisinger did the same in a well controlled DBT at 20kHz and got null results.
Aliens use advanced MRI technology to prove 100kHz audibility.
- Status
- Not open for further replies.