New means of distortions' measurement (Weighted Distortions)

[JohnRoberts]

Perhaps I have a personal problem but after decades designing audio gear I have learned to not trust poorly designed listening tests (including my own), just too many ways to get it wrong. I find uncontrolled listening tests good for later in the design stage to check for something unanticipated and unexplored that wasn't metered.

Test equipment is only as good as the operator's experience and judgment. I'm still working on mine.

I find the esoteric audiophile community a poor source of scientific advances but there are rare exceptions. Jon Risch, the guy with the multi-tone IMD test is unquestionably an audiophile (trust me). So sometimes there's an intersection of that pursuit and actual useful engineering, while he has his share of detractors for some of his suggestions, I've found him mostly credible and he hasn't lost touch with engineering fundamentals.

JR

 

[Guest]

Dominant of some representative system is not a problem, it is the common case among human beings.

Linguists discovered long time ago that different people use different predicates. Some people more often use predicates that indicate visual information, they often have high voice and high gestures; some people use audial predicates and their voice is smooth, they often keep a hand between their mouse and ear to hear own voice better; some people prefer words of tactile senses, they often move slowly, keep hands low during gesticulations. Even the same person often prefers different representative systems depending of state of mind.

The problem is, when a reasoning interferes with perception in the way, "what I perceive is wrong because I know this is impossible".

My point is, if so many audiophiles are not satisfied by measurement systems it does not mean they all are fools or half deaf, it means that we do not measure right so have to continue working on the subject.

That's why I've started the topic. I realized that many measurement techniques were used to address particular problems of particular technologies, and no one of them can be used to compare apples VS oranges, in terms of the end result for listener's satisfaction. My 30 years old P x THD measure was completely wrong; I used it to convince myself that my transistor amplifiers were better than tube amplifiers of that era, they offered less THD on a big power, but I then discovered that I was wrong very soon and gave up the idea... Especially, when I was impressed by A+C class amplifier that sounded great, but I did not believe own ears, neither ears of others, thinking that I was wrong implementing satisfactory result cheap, I was thinking that it was kind of cheating... I never repeated that design until later discovered an article in Wireless World about "Current Dumping"... But the article was tricky and again suggested low THD describing some kind of "bridge" that magically made 2 non-linear functions to continue each other without additional non-linearities... In reality, crossover point was shifted on higher power, but it was still the crossover point! An ordinary "properly biased" amplifier made of the same parts consuming the same idle power would have similar THD, but sounded less pleasant!

I believe, like F&M proved the concept with help of 11 experts only (thank you SonsOfThunder!), something similar can be done with measurement of impact of errors on perception of naturalness of a sound reproduction.

 

[JohnRoberts]

My apologies to the list for this dialog, but perhaps you're looking under the wrong rock for answers.

IMO musical playback is critically flawed due to attempting to capture, store, then reproduce a phenomenon in a media of less dimensions (gross oversimplification) than the actual event.

This general sense that something isn't right or could be better may drive some to an almost fanatical pursuit of perceived improvements that may sound better in some circumstances but may be tantamount to rearranging the lumps under the dimensional rug.

I don't mean to dismiss the pursuit of better sound quality but submit that within the dimensions we store and reproduce we have linearity well in hand. Speakers somewhat less so and there's some interesting work going on there, but speakers are also at the point of the dimensional disconnect and ripe for modification and experimentation.

I am not optimistic there is some as yet undiscovered, unmeasured, non-linearity that will trump the larger problems with realistic natural reproduction. I found that when I had a live band set up in my living room, it sounded exactly like a live band, in my living room. When they left it no longer sounded like a live band in my living room.

JR

PS: Large numbers of people can be of one mind and still be wrong. Just look at elections.

 

[Guest]

Interesting information:

curves of equal loudness from Robertson and Datson differ from curves of Fletcher and Munson!

(D W Robinson and R S Dadson «A re-determination of the equal-loudness relations for pure tones» in Br. J. Appl. Phys. 7, 166—181 ,1956).

They used a speaker in well damped room.

Fletcher and Munson used headphones!

«Loudness, its definition, measurement and calculation» in J.Acoust. Soc Am.5, 82-108 (1933).

I may expect good speakers in 1956, but good headphones in 1933... Hmmmm....

 

[mikep]

you know what really works, as far as listening tests? live with something for a while, use it every day, like professionally. then try comparing it to something else. you will have a different perspective, maybe less emphasis on some of the good aspects of the sound (wow factor or whatever), and you hear some of the things the DUT is *not* doing, both good and bad. I think audio hindsight is 20/20. not very time efficient way to evaluate new topologies, though.

mike

 

[Guest]

[quote author="mikep"]you know what really works, as far as listening tests? live with something for a while, use it every day, like professionally. then try comparing it to something else. you will have a different perspective, maybe less emphasis on some of the good aspects of the sound (wow factor or whatever), and you hear some of the things the DUT is *not* doing, both good and bad. I think audio hindsight is 20/20. not very time efficient way to evaluate new topologies, though.
[/quote]

Also, I remember a question of one CEO of one automotive manufacturing corporation to a sales rep of one computer company, "Why you don't use your own computers for your own manufacturing if they are so good for me?"

So, I always use my concert systems at home for music and movies. Right now I hear as if a real symphonic orchestra is playing downstairs, and as if my living room is of the size of a big philharmonics hall