R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1] 2 3 ... 10   Go Down

Author Topic: the "high frequency transients" fallacy  (Read 53941 times)

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
the "high frequency transients" fallacy
« on: February 25, 2005, 07:18:25 PM »

Listning to transients:

It is a well accepted fact that even a very good ear does not hear much above 20KHz, that most mics do not pick up energy much above 20KHz, and that most speakers will not generate sound much above 20KHz. It is also well known that most of the energy generated by musical instruments is at frequencies not much higher then 20KHz.

For some reason, many ear types seem to have bought into the notion that all that I have stated above is true for some of the music, and there is also another part of the signal which they refer to as “high frequency transients”.

The concept of transient energy is well accepted by all EE’s (not just in audio) to differentiate between signals made of repetitive cycles and signals that have little or no repetition. EE's and math types make the distinction between steady state and transients (non steady state) because the math and analysis of steady state signals is simpler and easier. Non steady state analysis (transient analysis) requires much more complicated tools.

However, the non repetitive audio signal behaviour (transients) does NOT contain frequencies higher than audio. Non-repetitive behaviour DOES NOT require, imply or call for high frequencies.
The fast attack of sound, be it drum, bell or a muted trumpet may have some low energy at frequencies above human hearing. But if your system could pick 100KHz transient energy, it would certainly be able to pick up and react to 100KHz sine wave! Why bother with 100KHz energy? If you can not hear 30Khz sine wave, you can not hear 30khz transients.

The whole concept of “high frequency transient” has been repeated over and over numerous times in the audio industry, in audio magazines, in marketing material and by people intend on finding "some sort of an explanation" for various fallacies they are trying to explain or promote.

After so much repetition of that faulty concept, I would not be surprised if many ear types will simply refuse to hear what I am saying but I am stating the FACTS!

Example – “multiple choice test question”:
Which case yields the fastest voltage change you can archive with say a 22KHz bandwidth system (such as a 22KHz mic, ear or speaker)?  I am talking about the steepest slop, thus the fastest changing signal.

A. 100Hz 1V peak square wave
B. 1KHz 1V peak square wave  
C. 5KHz 1V peak square wave
D. 10KHz 1V peak square wave
E. 22KHz 1V peak sine wave

The 5 waves are plotted below:

index.php/fa/719/0/

A. 100Hz 1V peak square wave is the red trace
B. 1KHz 1V peak square wave is the blue trace
C. 5KHz 1V peak square wave is the green trace
D. 10KHz 1V peak square wave is the purple trace
E. 22KHz 1V peak sine wave is the black trace

The fastest signal slope you can ever get in a 22KHz bandwidth system is not a transient. It is the slope of a steady state tone sine wave at the edge of the bandwidth - it is a 22KHz sine wave. The fastest signal in a 30Khz bandwidth system is a 30KHz sine wave, and so on. It is faster then the fastest transient you can create within the given bandwidth!

I know that many in audio “became attached” to that MISCONCEPTION about high frequency transients. Sorry to burst the bubble, but it is long overdue to have that nonsense cleared up. It is time to refocus on the activity to what happens WITHIN the audio band, and put an end to that fantasy about things that do not matter and things that do not exist.  

If you want to test your system or your ear for "the fastest signal it can handle", go for the highest sine wave frequency within the available bandwidth. That is true for ears as well as speakers, amps and everything else.

There is no such a thing as high frequency transients extending beyond the audio bandwidth. A mic that can not capture a steady state signal above say 20KHz, can not capture transients above 20KHz. The same is true for the ear, speaker or anything else in the system. We can argue about how high an ear hears, and be sure it never gets higher then 30KHz. For most people it is well below 20KHz, be it steady state or transient energy.
 
Regards
Dan Lavry
www.lavryengineering.com
Logged

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: the "high frequency transients" fallacy
« Reply #1 on: February 25, 2005, 11:19:51 PM »

danlavry wrote on Fri, 25 February 2005 19:18

Listening to transients:
......

There are no such things as "high frequency transients" which extend beyond the audio bandwidth.

A mic that cannot capture a steady state signal above 20KHz cannot capture transients above 20KHz.

This is true for the ear, a speaker or anything else in the system.
 





This paragraph should be printed on T-shirts and stickers! I'll happily wear one at the next AES.... (pardon my slight edit for grammar...)

Joe Bryan

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: the "high frequency transients" fallacy
« Reply #2 on: February 26, 2005, 03:01:44 PM »

Dan makes a valid point given the assumptions he's made, specifically, that the system is band-limited to 22kHz. However, the issue of transient response is more complicated than he's presented it.

A perfectly band-limited environment only exists inside a digital system, and is nothing but a theoretical framework. Bringing this virtual representation into perceivable reality requires going through analog systems which are *not* linear, and which *are* affected by high frequency transients, although not in the way it's been presented.

For example, analog circuits commonly use lots of negative feedback to counteract their inherent nonlinearities. The corrective nature of this feedback requires wide open loop response, or the feedback stops being corrective and starts being another source of distortion. If the feedback wraps around multiple intermediate stages without local feedback, these intermediate stages can go very non-linear under certain conditions.

The simplest way to make these circuits behave correctly is to make their frequency response as wide as possible without being overly sensitive to high-frequency interference. This bandwidth tradeoff is a big issue in audio circuit design.

One approach is to use less negative feedback with lower open loop gain. This approach trades a slight increase in low-order (musical) distortion for less high-order (non-musical) distortion, and has the nice side-affect of producing a circuit that has a gradual increase in distortion instead of a rapid increase at the clipping point. A lot of excellent pro-audio gear uses this approach.

High-order distortion components are especially audible when the fundamentals are low frequency, and the harmonics are still within the audio band. Low frequency signals are typically louder than high frequency signals (because of the ear's reduced sensitivity to low freq), so their harmonics rapidly become non-musical (inharmonics) and are less likely to be masked by higher frequency musical signals.

When multiple signals are present, their high-order harmonics can interact to produce sum and difference signals that add even more noise to the signal. Most importantly, difference signals between non-audible distortion components (>30kHz) can appear in the audio band.

Lowering the analog system's frequency response (band-limiting it) doesn't get rid of these signals because the negative feedback stops being corrective and you get more high-order harmonics, especially in intermediate amp stages inside the loop.

Another issue that's related to bandwidth is phase linearity. Wide-band systems have a flatter audio band phase response, and are thus more linear. There's no way to lower an analog system's bandwidth without affecting phase linearity and/or amplitude response ripple.

In conclusion, just because you can't hear certain signals doesn't mean the system doesn't have to handle them correctly.

-Joe

Joe Bryan
VP Engineering
Universal Audio
Santa Cruz
Logged

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: the "high frequency transients" fallacy
« Reply #3 on: February 27, 2005, 03:09:56 PM »

What does any of what you said have to do with my statement?
Your post can be summarised as "lets talk about other things".  

You came in with all sorts of OFF SUBJECT comments, from intermode distortions, to amplifier open loop and close loop and so on. It only serves to change the subject and to focus away from the subject.

You conclude by saying:
In conclusion, just because you can't hear certain signals doesn't mean the system doesn't have to handle them correctly.

I disagree. The system has to handle what you hear correctly. Preventing intermod, keeping the phase linear and much more. Using wide band amplifiers, making sure you do not have off band signals capable of mucking the process are all means to an end. It is the designer headache. It does not change the goal from "good performance audio" into a "wide band system". The goal is "good performance audio", whichever way you get there!

Mind you that much of that high frequency signals you talk about having to Handel, are not there to start with. Having to handle intermod distortions is not a goal. It is a problem that occured in the electronics and it is important to prevent it or eliminate it, but that too is totally off subject!

You talk about the need to handle those signals above audio, that most mics do not pick up, and the ear can not hear. How did the signals get there? You talk about methods to prevent generation of unwanted signals – distortion. We are not talking about how to design linear low distortion gear. We are not talking about the need to prevent unwanted high frequency distortions from occurring. We are talking about people believing in the existence of musical high frequency transients! At frequencies above “the rest of the music”. That is, before it even hit the mic!

I do not have a problem with a designer opening the bandwidth of say a power amp to 100KHz as MEANS to getting good phase linearity. But the goal IS NOT to have 100KHz bandwidth. The goal IS to have good 20KHz (or slightly above it). It is therefore a dis service to sell wide bandwidth as a goal for good audio. The goal is dictated by what you hear, the rest is just means to an end.

My statement is: we can hear steady state (repetitive cycles) tones, we can hear transient signals (non repeating signals). We can hear music which is a combination of transient and non transient energy. The claims that we can hear high frequency transients beyond the frequency that makes the rest of the music are wrong!

Please stay on subject. I will be glad to talk about negative feedback or phase linearity in another thread dedicated to that. But negative feedback, linearity, intermod apply to steady state and transient signals... Why the distinction?

This thread is about the fallacy of  breaking what one hears into 2 parts – the transient part and the “regular” (non transient) part, and falling for a wrong argument that we can hear higher frequency transients than the “ordinary non transient” part.
I see endless reference to such BS being spread around.

BTW
I would be glad to, at a later date, talk about opening  the amplifier bandwidth, feedback and so on.But please, limit your responses in this thread to the subject at hand.

Regards
Dan Lavry
www.lavryengineering.com
Logged

Joe Bryan

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: the "high frequency transients" fallacy
« Reply #4 on: February 27, 2005, 06:32:57 PM »

Dan-

I'm sorry you missed the point of my discussion about nonlinearities and how this relates to in-band perception. Put simply, real audio gear is non-linear and the harmonics produced by these nonlinearities are the source of the mysterious high frequency content you say doesn't exist. The interaction of these distortion components manifests itself in many ways (internal distortion in multistage amps with nonlocal feedback, intermod into the audio band, crest factor effects in dynamics detectors), and these effects cannot be ignored.

Specifically, I pointed out that because the theoretical framework you use to back your argument does not exist in a perceptual reality, and does not take into account the nonlinear effects (whether they're desireable or not) of real systems, it is not the final word on this subject.

I really appreciate your efforts to educate people in the technical fundamentals in this forum, and I've forwarded many of your posts to non-technical types because they're factual and well presented.

However, I'm debating you on this topic because even though you correctly point out that "high frequency transient response" is a misnomer, you do not allow that the concept it tries to address is valid, i.e., that it's really an attempt to describe how a system handles distortion, and this oversight cannot be ignored because it's one of the most important topics in our field.

-Joe
Logged

David Satz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 661
Re: the "high frequency transients" fallacy
« Reply #5 on: February 27, 2005, 09:25:39 PM »

Joe, I see your viewpoint as well as Dan's, and I understand why you mentioned the points which Dan said were off-topic. His initial statement was simply that in a band-limited system, no transient signal can have any greater maximum slew rate than a full-scale sine wave at the Nyquist limit. He was talking only about the rate of change in the signals themselves.

You responded by describing some well understood, real-world situations in which circuits with wider bandwidth and higher slew rate could be advantageous--though that wasn't disputed in what Dan wrote.

Unfortunately, nonmathematical general statements can be highly misleading, and are commonly exploited for commercial gain. Years of out-of-context statements in the audiophile press have convinced legions of audiophiles that all op-amp circuits (indeed all feedback amplifiers whatsoever) are sonically tainted due to phase shift that leads inexorably to dynamic distortion, regardless of context or amount.

Such statements rapidly become nonsense, because not all distortion has equal audible importance. There is some distortion even in a class-A circuit, even in simple passive components--even in wires. But can one hear it? That depends on the specific case. Thus merely pointing out potential distortion mechanisms says nothing useful. One must design consciously, measure wisely, and listen carefully, of course, but I would hope that we all knew that already.

I'd like to bring attention back to Dan's point about signals themselves. Jangling keys, "bongers" and exotic, close-miked percussion may offer useful tests of components or microphones--but not necessarily for the reasons people have been taught to believe. Spefically it's neither because they're "faster" than a full-range sine wave at the highest frequency of interest, nor because they contain higher-frequency signal components. Such signal components can't cause distortion if they've been filtered out before they reach the device under test! And in that case, the signal can't exceed the slew rate of Dan's full-scale sine wave, either.

Would you agree?

--best regards
Logged

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: the "high frequency transients" fallacy
« Reply #6 on: February 28, 2005, 12:12:41 PM »

Joe Bryan wrote on Sun, 27 February 2005 23:32

Dan-

I'm sorry you missed the point of my discussion about nonlinearities...

Specifically, I pointed out that because the theoretical framework you use to back your argument does not exist in a perceptual reality...

-Joe




Hello Joe,

My statement IS ABOUT HEARING (How high the ear hears). It has nothing to do with electronics. My statement does not call for any discussion about how electronic gear handles high frequency or low frequency. My statement does not call for arguments about negative feedback or intermod distortion. My statement is about a misconception that has been spread around in order to sell gear. It has nothing to do with any specific electronics gear.

Imagine yourself in the year 1800, listening to a concert in Paris. Or to a tribal ceremony in Africa… There is no electronics involved, just you listening to music. My statement is:
It is groundless and wrong to “break the sound” you hear into 2 compartments of different frequency bandwidth – the “transient compartment” and the “rest of the music, the non transient part”.

I would like everyone to understand that. When we all do, much good will happen to those that are trying to make an honest living in the audio industry. That “notion” of hearing “high frequency transients beyond the normal audio” is a lie that has been propagated throughout the industry for a long time. It is being used as a tool for selling all sorts of audio gear (even simple copper wire) and for promoting concepts such as that 192KHz sampling rate for audio.

The first step to sell someone a piece of gear that can “pass a signal” to say 50KHz, is to convince the buyer they can hear 50KHz. It is easier to market using "feel 50KHz".  

I agree with some of your comments about feedback, distortions and more. I also disagree with some. Given that feedback, intermod  and other comments you made apply to both steady state and transient, I found a number of your comments to be off the subject. I just do not want to get into a discussion about design detail here, not until we can talk about the same issue.

It would be good for all to realize what transient means and what steady state means. Steady state – a sound made of repetitive identical cycles, is very useful in the lab for testing gear. One can loosely say that a long note (say an organ key) has a “lot of steady state” in it. Between the time the sound builds up (the attack) and the time the sound stops, there is a “constant tone”, a bunch of repetitive (or nearly identical) cycles. We can call it “quasi steady state”. But as a rule, music is made out of a lot of time varying sounds, and all of the variations are transient. And it all (steady state and transient energy) occupies the same bandwidth. The transients do not occupy higher frequencies (in fact, the highest frequency in a given bandwidth belongs to a periodic signal at the end of that bandwidth).

So one does not need to buy gear that extends to those higher frequencies. There are times where designing electronics gear with some extra bandwidth may be a good strategy for accommodating  the bandwidth we hear. I too use that “tool”. But it is not about enabling the ear to hear energy at higher frequencies.    

Regards
Dan Lavry
www.lavryengineering.com



Logged

Joe Bryan

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: the "high frequency transients" fallacy
« Reply #7 on: March 01, 2005, 12:54:11 AM »

Hi Dan-

I'm not quite ready to accept your assumption that the ear has the band-limited reponse your argument requires.

I'm not an expert on human perceptual physiology, so I'll leave this point for others to debate. However, I know how circuits behave, and if the physical issues I've mentioned have analogues in our hearing system, I wouldn't be very surprised.

In the mean time, I'll continue to believe my own ears when it comes to audio quality issues.

Cheers,

-Joe
Logged

Sahib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 429
Re: the "high frequency transients" fallacy
« Reply #8 on: March 01, 2005, 06:18:45 AM »

Joe,

I can see your point of the impact that in-audible frequencies may have on human being. For example an ultra high frequency shock pulses (such as weapons) directed at a person may not be heard but would have an effect perhaps physcologically at the least. Equally you may have an argument that a second and third harmonics of say 20kHz will not be audible but may have an impact on our perception of the music. And that is where the problem lies. It becomes completely subjective and impossible to prove. Besides even if the equipment can not handle those harmonics, who says that one's physcological condition at the time of listening to the music can not alter his/her perception of 20kHz. So, unfortunately in a situation like that a snake oil man steps in and sells the person an amplifier that handles 50kHz. My opinion is, it's the emperor's clothes, it's not there but I respect your views.

Regards,
Cemal
Logged

RobertRandolph

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 54
Re: the "high frequency transients" fallacy
« Reply #9 on: March 01, 2005, 08:24:39 AM »

I can say by all hearing tests Ive been subject to by licenses audiologist give me an average ability to hear.

BUT. I have done blind tests with monitors (specifically earthworks) that can pass frequencies above 20khz. Blind folded, and simply passing fullscale 25khz and 28khz signals at 96khz sample rate with apogee converters (not at my studio). Monitor sysetm was calibrated to the k-system  k-14.

Every single person was able to tell when the signal was on. And I can describe it very easily as simply feeling a change in the air. I cant claim to have "heard" it, but I know that of the 4 people there who participated in the test there was not a single incorrect answer... I dont recall the exact number of repititions but it was above 100.

I cant say for sure the result was due to resonances from the speaker mechanism in the audible band or simply the ability to feel the minute change in pressure by some human mechanism. Either way I have been convinced more than once that in many modern listening systems, frequencies above the audible range have an impact on the listener. Perhaps by imperfect listening systems, or by human ability. Regardless it exists.

And I invite anyone to try it.
Logged
Now I know a disease that these doctors can’t treat
You contract on the day you accept all you see
Is a mirror, and a mirror is all it can be
A reflection of something we’re missing

Sahib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 429
Re: the "high frequency transients" fallacy
« Reply #10 on: March 01, 2005, 09:41:51 AM »

Robert,

I can not disagree with that but my point would be that what sort of qualities that sensation would bring into preceiving the music piece as a whole. Again, that's where we are hitting the danger of having to accommodate the snake oil.

Regards,
Cemal
Logged

RobertRandolph

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 54
Re: the "high frequency transients" fallacy
« Reply #11 on: March 01, 2005, 10:31:28 AM »

Sahib wrote on Tue, 01 March 2005 09:41

Robert,

I can not disagree with that but my point would be that what sort of qualities that sensation would bring into preceiving the music piece as a whole. Again, that's where we are hitting the danger of having to accommodate the snake oil.

Regards,
Cemal


Agreed. But I think it is dangerous to assume that humans cant perceive frequencies above 20khz in any way.
Logged
Now I know a disease that these doctors can’t treat
You contract on the day you accept all you see
Is a mirror, and a mirror is all it can be
A reflection of something we’re missing

Sahib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 429
Re: the "high frequency transients" fallacy
« Reply #12 on: March 01, 2005, 11:48:46 AM »

I promise this wil not go on forever but perceiving a sound above 20kHz does not constitute 'hearing' it but it constitutes 'feeling' it and that is where we have to draw the line.  Otherwise you can also feel a UHF blast of shock waves and how do you include that in the equation.

Saying 'I can hear above 20kHz ' is simply not good enough. It has to be measurable and if it is measured than we would perhaps go back and re-establish the facts in audio band. We have done this before with the flow of electricity after Benjamin Franklin's wrong assumption but until it is proven we have to stick to the existing facts and figures.

Regards,
Cemal
Logged

RobertRandolph

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 54
Re: the "high frequency transients" fallacy
« Reply #13 on: March 01, 2005, 12:00:52 PM »

I dont think anyone claimed we could hear sound above 20khz, but I strongly beleive (and I think it is easily verifiable if it hasnt been already... I havent researched it), that we can determine the presence of that material through some other means. Even if this mean is the ears and we are not able to fully discern the material as intelligable.

Perhaps such a presense is why many people feel wider bandwidth listening systems sound more "Real" simply because of the presence of sound they are used to being able to perceive in the "real world".

In the end, I personally work at 44.1/24 98% of the time. Ive never had a client or listener complain the high end wasnt pure enough... let alone a musician (Who's ears are normally battered to death and brains beaten by drugs and alcohol lol Smile j/k)

Logged
Now I know a disease that these doctors can’t treat
You contract on the day you accept all you see
Is a mirror, and a mirror is all it can be
A reflection of something we’re missing

David Satz

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 661
Re: the "high frequency transients" fallacy
« Reply #14 on: March 01, 2005, 12:41:42 PM »

Robert, you wrote regarding a group listening test in which everyone reacted to the presence of signals above 20 kHz, "I cant say for sure the result was due to resonances from the speaker mechanism in the audible band or simply the ability to feel the minute change in pressure by some human mechanism. Either way I have been convinced more than once that in many modern listening systems, frequencies above the audible range have an impact on the listener. Perhaps by imperfect listening systems, or by human ability. Regardless it exists."

This is the kind of thing that really, really irks me about so many discussions in audio. Before any further conclusions can be drawn from the experience which you report, someone needs to put up a microphone and a spectrum analyzer and find out whether audible tones were accidentally being generated at frequencies below 20 kHz. Then people can start forming sensible hypotheses--but not before.

Until that is looked into, the possible implications of the experience are so diverse, they're all over the map. Maybe loudspeakers need to be better; maybe all power amplifiers should contain brick-wall filters; maybe the D/A converters were distorting (regardless of how prestigious the brand name); maybe ... well, six other things I can think of. But I'm sure that you can think of those six other things just as well as I can.

Thus you should realize that this experience, without more evidence, doesn't lend weight to any particular conclusion whatsoever--unless you want to get cynical and list it as an example of the general decline in critical thinking.

--best regards
Logged
Pages: [1] 2 3 ... 10   Go Up
 

Site Hosted By Ashdown Technologies, Inc.

Page created in 0.084 seconds with 21 queries.