Measurements and the Correlation of What We Hear

[garylkoh]

It's there, I have heard it,it mainly effects resolution and the effects can be startling,no pun intended. Just to clarify (the current that goes through the wire).

I've heard it too, but I don't know how to measure it. In the Cable Dialectic, we went through 11 pages, and computer models and simulation, and couldn't come to a conclusion there. None of the simulations show that anything we did to a cable can be audible. But I hear it. I don't know how to measure it, nor how to simulate it, and there are then there will be those who will proclaim that it's my imagination and/or expectation bias.

 

[RogerD]

I've heard it too, but I don't know how to measure it. In the Cable Dialectic, we went through 11 pages, and computer models and simulation, and couldn't come to a conclusion there. None of the simulations show that anything we did to a cable can be audible. But I hear it. I don't know how to measure it, nor how to simulate it, and there are then there will be those who will proclaim that it's my imagination and/or expectation bias.

Hi Gary,

Neil Muncy has done tests on SCIN Shield current induced noise ( http://www.rane.com/note166.html ,but I have found that using a simple cable snap EMI/RFI filter on both ends of a PC (NBS), IC's,digital cable and speaker cable can have a very noticeable effect on clarity and soundstage. FWIW

 

[DonH50]

An NPR test would be interesting for showing low-level detail and how a given component handles wideband stress. It should be fairly easy to set up now in the digital age; in the past it was exceedingly painful due to the filters required.

 

[RogerD]

An NPR test would be interesting for showing low-level detail and how a given component handles wideband stress. It should be fairly easy to set up now in the digital age; in the past it was exceedingly painful due to the filters required.

NPR test? Is it this?

http://www.lintech.com/PDF/npr_wp.pdf

 

[DonH50]

Yes, sorry, should have defined the term... It was (is) used by the phone company and in many RF systems. It is a torture test for most circuts. The basic idea is to drive the device under test (DUT) with a wideband noise signal with a very narrow notch, then look what lands in the notch at the output. With DSP, we should be able to generate the noise signal, then sweep a narrow notch across the audio band and measure the results. It will not tell everything about the component but I bet it reveals a lot...

 

[garylkoh]

Thanks, Don. No, I have not tried a NPR test on the amplifier - will have to figure out how to set this up in the audio frequencies. The software I use doesn't have this as a feature

 

[DonH50]

Hi Gary,

I was just thinking out loud, didn't have you in mind explicitly. I swore to myself I was going to stay out of this thread, but oh well...

It is a fairly unusual test (for audio, anyway, and AFAIK), but I think it could be a really useful one, albeit likely to show weaknesses in a lot of designs. As a designer, I always dreaded it... I do not know if it would correlate, or how well, to listening tests, but if people are hearing low-level detail being lost in some systems, this test should show that.

No worries - Don

p.s. People might be surprised to see how their speakers measure distortion-wise... Sure to revive the "does THD matter in my amp" debate.

 

[Phelonious Ponk]

I've heard it too, but I don't know how to measure it. In the Cable Dialectic, we went through 11 pages, and computer models and simulation, and couldn't come to a conclusion there. None of the simulations show that anything we did to a cable can be audible. But I hear it. I don't know how to measure it, nor how to simulate it, and there are then there will be those who will proclaim that it's my imagination and/or expectation bias.

Were you able to rapidly switch between the cables? Was the output volume-matched? Was knowledge of which cable was playing when removed from the listening? I don't know what you heard, Gary, but if any of the above was not the case, you really can't be sure yourself.

Tim

 

[RogerD]

Were you able to rapidly switch between the cables? Was the output volume-matched? Was knowledge of which cable was playing when removed from the listening? I don't know what you heard, Gary, but if any of the above was not the case, you really can't be sure yourself.

Tim

Tim,

I can't speak for Gary, after spending 2000 hours with my system in 12 months........ now do I need to address this question really?

 

[microstrip]

(...) p.s. People might be surprised to see how their speakers measure distortion-wise... Sure to revive the "does THD matter in my amp" debate.

Don,

As distortion in loudspeakers is most times orders of magnitude higher than that of amplifiers we could imagine an experiment : measure the distortion of a typical loudspeaker and modify an amplifier to have the same distortion figures as this speaker and electronically equalize to compensate for FR. Then play it through a loudspeaker with very low distortion - e.g. Quad ESL63 . Would it sound like the original loudspeaker? Tests should be made in anechoic conditions, to avoid the contribution of the dispersion patterns.

 

[Phelonious Ponk]

Tim,

I can't speak for Gary, after spending 2000 hours with my system in 12 months........ now do I need to address this question really?

Nah...you don't need to address anything you don't want to address.

Tim

 

[microstrip]

Were you able to rapidly switch between the cables? Was the output volume-matched? Was knowledge of which cable was playing when removed from the listening? I don't know what you heard, Gary, but if any of the above was not the case, you really can't be sure yourself.

Tim

Although I understand your two last requirements, I can not understand why switching must be as rapid as a neutrino ...

 

[RogerD]

Nah...you don't need to address anything you don't want to address.

Tim

Gee.....I didn't think anything had changed

 

[garylkoh]

Were you able to rapidly switch between the cables? Was the output volume-matched? Was knowledge of which cable was playing when removed from the listening? I don't know what you heard, Gary, but if any of the above was not the case, you really can't be sure yourself.

Tim

No, Tim. I didn't do all that. I tried building a switch box to rapidly switch between cables, but the difference inserting that switch box was greater than the difference between the two cables. Unfortunately, I don't have a hope to being objective because with my expectation bias, I can even hear the difference when I remove the loudspeaker binding post nut (I'm using banana plugs).

 

[DonH50]

Don,

As distortion in loudspeakers is most times orders of magnitude higher than that of amplifiers we could imagine an experiment : measure the distortion of a typical loudspeaker and modify an amplifier to have the same distortion figures as this speaker and electronically equalize to compensate for FR. Then play it through a loudspeaker with very low distortion - e.g. Quad ESL63 . Would it sound like the original loudspeaker? Tests should be made in anechoic conditions, to avoid the contribution of the dispersion patterns.

Hmmm... Even in an anechoic chamber the radiation pattern matters so they would not sound exactly the same as you move around, but maybe perfectly on-axis it would work. It should confirm FR and distortion changes, at least. I am not sure about the (audibility of the) phase transfer function, but of course we should be able to match that as well.

Note even 'stats and other planers like my Maggies will have high distortion when overdriven and/or near the limits of their frequency response. And, I have read some astonishingly (to me) low distortion numbers for some conventional dynamic speakers (<0.5%).

Not sure this is relevant to the topic? Interesting, but off-topic, something that never happens, natch.

 

[garylkoh]

Hi Gary,

I was just thinking out loud, didn't have you in mind explicitly. I swore to myself I was going to stay out of this thread, but oh well...

It is a fairly unusual test (for audio, anyway, and AFAIK), but I think it could be a really useful one, albeit likely to show weaknesses in a lot of designs. As a designer, I always dreaded it... I do not know if it would correlate, or how well, to listening tests, but if people are hearing low-level detail being lost in some systems, this test should show that.

No worries - Don

p.s. People might be surprised to see how their speakers measure distortion-wise... Sure to revive the "does THD matter in my amp" debate.

Haha! This thread is tempting to jump into. I'm out of town this week, then there is the RMAF next, but I'll think about how to implement the NPR test in audio amplifiers when I get back.

Typical distortion figures for loudspeakers (which vary WILDLY depending on how you measure them) range from about 5% to over 50%. I've measured non-servo subwoofers to go up to 50% even below their distortion spl. The graphs look like a big smiley face, very high at low spl's (which I figure is caused by air not transmitting enough) to very high at large spl's (which I figure is air being pushed aside by the rapidly moving cone). The elasticity of air makes measurement of distortion very difficult for loudspeakers.

 

[Orb]

Orb, Thanks.

The problem with a real-world measurement like this is that it is difficult to compare the input and the output. The complexity of the input means that jitter in the measuring instrument would invalidate the result.

The only significant measurements I've seen in this area have been the experiments conducted by Nordost/Vertex - but because they use a PC and soundcard for the measurement, they have been discredited on this forum even though they say that they used math to match the input and the measured output.

All the studies I've read on this subject have been done with listening panels, and this is not the subject of this thread - which is measurements. As Frantz mentioned, these are bodily perceptions stimulated by some physical stimuli.... and if they are physical, there are many here who are sure that if it can be perceived, then they can be measured.

I'm of the camp "I can hear it, but *I* don't know how to measure it, and I would love for someone to tell me how".

The input and output can be done to compare IF done with the right tools, and they had the time - spoken to a few on this who have the skill and experience.
The problem is as I mentioned before, is that it will not be developed because it will not help in testing an engineered product in terms of linearity-performance-behaviour.
This test-measurement is only of benefit to understand the character of the actual real sound, which is to provide a very accurate way of seeing a specific measurement to what is heard.
If anyone can provide in this thread a way the discussion is going with other measurement parameters (along the more traditional engineering product lines) and how it can then be considered to show accurately real world implications on a complex sound...
I will be gobsmacked

BTW did you look at all the links I provided?
Did anyone take the time to read them
They provide clear indication on sound and perception and has nothing to do with bodily perception stimulated (it is about perceiving and translating sound), especially in terms of timbre, and show our understanding of how sound-instruments are perceived with measurements along the line I have mentioned.

As I mentioned earlier, this thread will reiterate endlessly because the focus is on the usual type measurements and not what is done for measuring sound-instruments in science papers relating to actually understanding-modelling what defines them, which includes looking at what makes complex subtle sounds different.

If a conclusion is to be reached, you need to be able to take those measurements most are discussing and then be able to correlate them to the complex waveforms and links I have provided so far.

With that said, if anyone has an interest regarding what I am saying send me a PM and I will try to find more papers/links that I have which touches on this subject (including more summary and easy to follow pages).
Probably makes more sense if I leave it at that for now.

Thanks
Orb

 

[Orb]

Orb

Notions like "musicality", "flow", "rightness", etc are our attempt to describe our aural perceptions with words, I will grant you, a difficult proposition. Those perceptions however, if they exist at all, are our body responses to physical thus measurable stimuli. Sounds exist in time and space and that all there is to it all measurable. I am somehow unsettled to see how some of us are bending around incredible loops to come up with the notion that something that exists (a stimulus which creates a perception) can't be measured ..i sincerely would have not thought this line of reasoning ..well.. reasonable but .. we are audiophiles ... we believe in stranger things ...
I will answer Gary post later .. It seems at first reading to belong in the "we-can-hear-it-but-can't-measure-it" category ... I disagree with that line of thought...

Just expanding upon musicality and my use of it that does make me wince everytime
I appreciate it is met with quite a bit of hostility by a fair few in audio forums (in general and not directed at you Frantz), also it seems to me to be a bit vague in how it can be interpreted.
But what is interesting, the term is used a lot by professional dance judges when evaluating dancers in a competition (talking about ballroom dancing), and it is used in classical music by many as well.

Not one I have read but maybe of interest to those who are musicians; has anyone read "The Anatomy of Musicality" by Rozalie Levant?
The author is a pianist, PHD in musicology, lecturer, and been director of music at the Bolshoi Drama Theatre.

Cheers
Orb

 

[fas42]

The basic idea is to drive the device under test (DUT) with a wideband noise signal with a very narrow notch, then look what lands in the notch at the output. With DSP, we should be able to generate the noise signal, then sweep a narrow notch across the audio band and measure the results. It will not tell everything about the component but I bet it reveals a lot...

Thanks for throwing that in, Don. There are myriads of ways of detecting how the system misbehaves apart from the normal AP tests, my thought bubble and this are just two. Another idea I've played with a few times as a thought experiment, which only requires ears, is to record someone with a very clear voice speaking a random sequence of digits, 1 to 10, for quite a decent length of time. Then mix that in with white, pink noise, whatever, with varying levels of noise compared with voice. Then playback at various volume levels and the task is obvious: correctly identify the numbers spoken.

Then you can apply any level of statistical hocus pocus to it that you like, it will all be very "scientific" ...

Frank

 

[fas42]

Typical distortion figures for loudspeakers (which vary WILDLY depending on how you measure them) range from about 5% to over 50%. I've measured non-servo subwoofers to go up to 50% even below their distortion spl. The graphs look like a big smiley face, very high at low spl's (which I figure is caused by air not transmitting enough) to very high at large spl's (which I figure is air being pushed aside by the rapidly moving cone). The elasticity of air makes measurement of distortion very difficult for loudspeakers.

I don't get this "speakers distort like crazy" business. Of course bass speakers have bad distortion but the ear is incredibly tolerant of this: you can mix in 100% distortion and people can't pick it.

Midrange and higher is where it's important, and over 20 years ago Martin Colloms had no trouble measuring speakers having 0.1% distortion or better in this key area, at reasonable listening levels. So what's all the fuss about??

Frank