Gedlee on distortion

[DDF]

Re: Gedlee on distortion

An extreme example might exist in comparing harmonic overtones of soprano saxophone and clarinet, both similar in size but different in shape resulting in different harmonic overtone spectra, producing very different tonal qualities on the same musical notes. The clarinet approximates a closed end tubular instrument with mostly odd order harmonics, very low levels of even order harmonics. The saxophone approximates a truncated conical instrument with mouthpiece near the apex, and exhibits 2nd order harmonics near equivalent in level to the fundamental, and 4th order harmonics not far down from the 3rd order harmonics, etc.

http://www.phy.mtu.edu/~suits/conecyl.html

http://www.phy.mtu.edu/~suits/clarinet.html

http://www.phy.mtu.edu/~suits/sax_sounds/index.html

Very interesting, thanks for the great info. A very plausible theory I've read in an auditory journal is that distortion audibility depends upon the waveform envelope. Theory goes that distortion audibility increases when the total signal envelope magnitude is at a low and distortion component may not be low. This mis alignment can be an artefact of group delay distortion for example.

The other thing that is important that I've seen auditory psychologists stress is that the level of audibility is a function of the spectrum seen at the inner ear. So its a function of the signal and its convlolution with the non linearities of the inner ear. For example, this means to apply masking effects.

I'll have to go back through my library but I can't recall reading any audition studies that found in general that even order distortion was much more benign than odd. I suspect that this may be an audio folk tale that somehow gained unchallenged legitimacy. It is true that 2nd order can mask the fundamental and vice versa but i think its quite situational.

I think Arnie took down his demo; Unfortunate as it was eye opening.

[DDF]

Re: Gedlee on distortion

This won't directly answer your question, but will give you an idea about the measurable differences:

http://www.linkwitzlab.com/mid_dist.htm

http://www.linkwitzlab.com/frontiers.htm#G

In the second link, note Linkwitz's comment, It seemed, that coming out of the loudspeaker a sound carried a comet like trail of newly created sounds with it, that formed a continuum of background sounds, which was completely absent from the live instruments and made the speaker immediately recognizable as such.

My personal observation is that low-distortion drivers allow me to hear better separation of instruments, voices etc., probably because there's less intermodulation. I've also heard the "newly created sounds" that Linkwitz describes, which to my ears are like a wall of fuzzy, ill-defined sound. That's especially true at higher SPLs. When there are many different tones present, the intermodulation products can become quite substantial.

I personally find phi multi tone approach helpfull where it injects numerous fundamental tones that are not harmonically related which allows the fundamentals to spectrally stay out of the way so its easy to see all the intermod products. I adapted this a bit and applied IEC spectral weighting to the fundamentals so that the amplitude of the fundamental frequencies follows the IEC spec for "average" music spectral density. This helps avoid pushing the high frequencies too hard.

The results arent easy to read in detail and relate back to audibility except in the grossest sense, but I have found this useful for identifying areas requiring attention. Its easy to see a higher nonlinearity quickly by increases in the "grass" (distortion products) between fundamentals. Another way to think about it is that it takes a distortion measurement and almost makes the results look like a noise floor increase due to the shear number of distortion frequencies created. It also gives a really handy qualitative assessment of what it might sound like as devices often respond by bunching up more distortion products within a particular frequency range and or having the products within a frequency range be notably louder than in other quency ranges

I was asked on the basslist once to review the gedlee approach and I probably still have that assessment somewhere. He's often publicly claimed thresholds of audibility of distortion which are much higher than I've read elsewhere and higher than those found by a subjective assessment sister group I used to work with (all of our testing was double blind and statistically significant but results were not public domain, being intellectual property)

[craigk]

Re: Gedlee on distortion

Very interesting, thanks for the great info. A very plausible theory I've read in an auditory journal is that distortion audibility depends upon the waveform envelope. Theory goes that distortion audibility increases when the total signal envelope magnitude is at a low and distortion component may not be low. This mis alignment can be an artefact of group delay distortion for example.

The other thing that is important that I've seen auditory psychologists stress is that the level of audibility is a function of the spectrum seen at the inner ear. So its a function of the signal and its convlolution with the non linearities of the inner ear. For example, this means to apply masking effects.

I'll have to go back through my library but I can't recall reading any audition studies that found in general that even order distortion was much more benign than odd. I suspect that this may be an audio folk tale that somehow gained unchallenged legitimacy. It is true that 2nd order can mask the fundamental and vice versa but i think its quite situational.

I think Arnie took down his demo; Unfortunate as it was eye opening.

I am amazed at how difficult some of you like to make things. go buy a driver with high 2 nd order distortion and a driver with high 3 rd and just listen. surely you can hear the difference. why do you have to find some white paper to justify every thing. cant you measure and listen and come up with your own conclusion.

[Deward Hastings]

Re: Gedlee on distortion

I can't recall reading any audition studies that found in general that even order distortion was much more benign than odd. I suspect that this may be an audio folk tale

Even order distortion is "on the octave" and always "in tune", and the second harmonic is close enough to the fundamental to be generally masked. Higher order even harmonics do effect instrumental timbre. Odd order harmonics may be dissonant, and even the third is seldom significantly masked. Odd order harmonics do effect instrumental timbre.

I think Arnie took down his demo; Unfortunate as it was eye opening.

It may have . . . um . . . not been (or demonstrated) quite what it purported to be (or to demonstrate).

[Deward Hastings]

Re: Gedlee on distortion

go buy a driver with high 2 nd order distortion and a driver with high 3 rd and just listen. surely you can hear the difference.

Too many other variables to draw any meaningful conclusion. If you were to take two otherwise identical drivers and wind the voice coil off center on one (to get 2nd order distortion, at least at high levels) and wind the other centered but with variable pitch (to get 3rd) such a comparison might be of some value . . .

[Pallas]

Gedlee on distortion

So are we to assume by your statement that the rest of us are the "wretches".

On can only assume from the above that you either didn't read the post to which you replied, or you didn't understand it. Scope is a useful thing to grasp.

;2077141]What snide condescension.

I agree that your quoted passage to follow is snide condescension. It also shows a lack of understanding of scope.

;2077175]But when he cites folks as "experts" who produce a product that is at best, mediocre (Geddes)***.

Your perception of the quality of someone else's product is out of the scope of a discussion of audibility of harmonic distortion, unless you wish to assert that said perception of the quality is due to the harmonic distortion profile of said product.

Furthermore, your perception of the quality of someone else's product is out of the scope the quality of that person's research into HD. Geddes has peer reviewed published scholarship on the subject. Do you?

And at any rate, had you read my posts for understanding instead of mere umbrage you would've noted that I quoted Toole to support my position, not Geddes. Indeed, I expressly stated that due to the narrow scope of devices to which they applied, Geddes quote from the OP does not apply to most home loudspeakers. The underlying truth may be the same, but Geddes was only talking about his speakers, and any application of his full quoted statement beyond speakers like his in the relevant aspects (big, high tech woofers and compression driver tweeters on WGs) is not reasonable.

[Deward Hastings]

Re: Gedlee on distortion

I personally find phi multi tone approach helpful

Agree. Unfortunately it's become associated with cable quackery (at least I think that's why it's generally ignored) and it doesn't lend itself to easy interpretation. It does, however, show more about the distortion we actually hear and why we hear it than any other available test, and it more easily generalizes to what actually happens with (complex) music signal . . .

[rickcraig]

Re: Gedlee on distortion

In defense of Pallas I met with him over dinner not long ago followed by a trip to do some listening at his house. Our discussion was enjoyable and he's obviously very knowledgeable and didn't come off as arrogant or offensive. Like Earl Geddes and myself he has strong opinions and preferences concerning speakers based on his personal experiences. I also met Earl once at an audio show and we had a good discussion, not a very long one, but distortion was one of the topics.

[carlspeak]

Re: Gedlee on distortion

The term 'masking' has been mentioned a few times here related to the distortion discussion. Masking is cited by Toole (Sound Reproduction.... 2008), Chap. 19, pg 452 as the basis for his conclusion that traditional measures of HD and IMD are almost meaningless. He goes on further to write that those measures do not quantify distortion in a way that can reliably predict a human response while listening to music or movies because those measures ignore the human ear, a known nonlinear device. At over 500 pages, Toole spent very little ink on the subject of distortion. It's not even listed in the Index.

What has become obvious to me, at least with a normally operating loudspeaker in good working order, is that distortion doesn't become annoyingly audible until a speaker is driven quite loudly, thus indicating it has drifted into non-linear operation. OTOH, with large venue professional speakers, this issue is more of a concern because they operate typically at high SPL levels. Voishivillo, of JBL has done considerable work in studying distortion. I always thought someone besides Toole and Olive must have paid considerable attention to this because much of JBL's pro audio business has to perform well at elevated SPL levels. Here is an ALMA 2009 presentation by Voishivillo on distortion worth a look... www.almainternational.org/yahoo_site_admin/assets/docs/voishvillo_-_nonlinearity_and_acoustics.106173319.pdf

Lastly, the referenced Chap. 19 in Toole's book summarizes the distortion discussion by stating that, in general, distortion in consumer loudspeakers has not been identified as a factor in overall subjective ratings. It may be there and measurable (as us DIY'rs do), but is low enough to not be an obvious factor in judgments of sound quality at normal foreground listening levels.

If one wants to delve into nonlinear distortion testing, perhaps consideration should be given to NRC's adoption of linearity testing (see http://techtalk.parts-express.com/sh...r-or-all-of-em). Test methodology is quite vague at their site. I wrote to them asking for more detail a week ago and have yet to hear back.

[carlspeak]

Re: Gedlee on distortion

In defense of Pallas I met with him over dinner not long ago followed by a trip to do some listening at his house. Our discussion was enjoyable and he's obviously very knowledgeable and didn't come off as arrogant or offensive. Like Earl Geddes and myself he has strong opinions and preferences concerning speakers based on his personal experiences. I also met Earl once at an audio show and we had a good discussion, not a very long one, but distortion was one of the topics.

So, did Pallas' speakers exhibit any DIY effort? or where they of commercial origin?

[DDF]

Re: Gedlee on distortion

I am amazed at how difficult some of you like to make things. go buy a driver with high 2 nd order distortion and a driver with high 3 rd and just listen. surely you can hear the difference. why do you have to find some white paper to justify every thing. cant you measure and listen and come up with your own conclusion.

I think this was uncalled for. I certainly do measure and listen. See my phitone post, and my comment about even order audibility came from a personal controlled test azi had already said

I think odd order can sound worse but ive also heard even sound worse. Jrt posted examples that illustrated when this may occur

I was trying to be helpful by posting knowledge off the beaten path. In this quorum, this is often met with open hostility vs appreciation. It provides an unpleasant impression of closed minded anger where anything that challenges the status quo is attacked not on the merits of the idea, but personally against the character of the poster

[Deward Hastings]

Re: Gedlee on distortion

Masking is cited by Toole (Sound Reproduction.... 2008), Chap. 19, pg 452 as the basis for his conclusion that traditional measures of HD and IMD are almost meaningless. He goes on further to write that those measures do not quantify distortion in a way that can reliably predict a human response while listening to music or movies because those measures ignore the human ear, a known nonlinear device.

Also by Geddes (Audio Transducers . . . 2002), Chap. 10, pg 237-241 and Chap. 13 pg 286-289, with the same conclusion as Toole. Geddes also makes a clear (and, IMO, correct) distinction between distortion in mechanical systems (speaker drivers) and distortion in electronics (where some non-linearities, most notably zero-crossing distortion, become more pronounced and noticeable at low levels).

[andy19191]

Re: Gedlee on distortion

The nature of non-linear as opposed to linear is that small things can have large effects (e.g. if a butterfly flaps it wings...)

That is incorrect. As non-linear distortion tends toward "small" it tends toward inaudible. We do not hear butterfly wings flapping (bird wings yes, butterfly wings no).

If the flapping of those butterfly wings causes a tornado that blows your house down then you might hear that but I think you may have misunderstood the point I was trying to make. If a system is non-linear then small differences can trigger large changes that can be significantly different from each other. This doesn't happen with linear systems.

What we have got at present works well so long as the reader understands what is and is not included in the measure.

Most don't. It appears that you don't.

Oh well, perhaps the thread was always doomed.

[Deward Hastings]

Re: Gedlee on distortion

It's a SEOS 12 using B&C DE250 . . . something about the midrange is "just right" in this one.

That's the combination I've settled on (over a W22 on a large baffle) for my HT "center". It gives the polar (and freedom from early reflections) that I want, excellent "clarity" and localization, and just sounds "right" for most if not all movie sound tracks (my room-filling dipoles take care of the rest, both movies and "music"). In the past I've tended to dismiss horns (and I'm still not ready for an "all horn" front wall) . . . but . . . per Mr. Natural, "pick the right tool for the job" . . .

[DDF]

Re: Gedlee on distortion

Here is an ALMA 2009 presentation by Voishivillo on distortion worth a look... www.almainternational.org/yahoo_site_admin/assets/docs/voishvillo_-_nonlinearity_and_acoustics.106173319.pdf

Thanks for posting, that was a great read. I pretty much agree with everything in it and it aligns with what I've read as well (my biggest beef with teh Geddes approach was also that it ignored the transform from external ear to inner cochlea). My background was in telecom audio, and that industry was adopting a PEAQ approach, but (as applied there at least) wasn't sensitive enough to catch the subtle differences audio enthusiasts care about in a hifi audio application.

If MP3 encoding can remove signal with varying levels of audibility (from low to high bit rate), I always felt it would be possible to use perceptual coding techniques to do the opposite: evaluate what is not masked by a system that subtracts and/or adds to the audio signal. The results could be evaluated in quantum steps each with a different confidence level of audible degradation in the same way that different bit rate encoders allow the possibility for different risk of audible degradation. ie
Run algo A which is relatively insensitive but has 95% confidence that a fail is audible; algo B is more sensitive but a fail may only have 75% chance of being audible; etc etc