It doesn't really matter what something 'sounds' like, if it does not sound like a natural version of the real thing, then the chain of reproduction is flawed. It almost impossible to find a 'perfect' source or thru-path, so we have to take it in steps.
We try to get the best sources and pieces of the chain that we can. Then we MAY try to change one of the pieces in the chain and try to hear for a difference.
Sometimes, one device might be 'clearer', another might be 'smoother', yet another may sound 'natural', yet give one a headache after a time. What one can hope to achieve is that ONE link in the chain is virtually 'invisible' and can't easily be bettered, without the rest of the chain being brought up to the same standard.
In my system, I had 4558's for years, in a Sony SACD player that I got perhaps a decade ago. It was considered one of the best available at the time. Yet, while I could listen to DVD and SACD through it, I could not tell any difference between DVD and SACD, and I presumed that it was the 4558's. Today, I have a unit with lm4562's and to me, it sounds pretty good, but I could probably do better with AD825's, or better yet, discretes.
This is what I do for a living, and I seem to be doing well enough at it to keep employed. That is enough for me.
We try to get the best sources and pieces of the chain that we can. Then we MAY try to change one of the pieces in the chain and try to hear for a difference.
Sometimes, one device might be 'clearer', another might be 'smoother', yet another may sound 'natural', yet give one a headache after a time. What one can hope to achieve is that ONE link in the chain is virtually 'invisible' and can't easily be bettered, without the rest of the chain being brought up to the same standard.
In my system, I had 4558's for years, in a Sony SACD player that I got perhaps a decade ago. It was considered one of the best available at the time. Yet, while I could listen to DVD and SACD through it, I could not tell any difference between DVD and SACD, and I presumed that it was the 4558's. Today, I have a unit with lm4562's and to me, it sounds pretty good, but I could probably do better with AD825's, or better yet, discretes.
This is what I do for a living, and I seem to be doing well enough at it to keep employed. That is enough for me.
Frank,
You are talking about scientific bias here. Yes you can set up an experiment to prove whatever you may be biased to get as a resultant. That is often the problem with so called experimental outcome if you have a per-existing notion of the outcome. Very hard to overcome without third party intervention that has no stake in the outcome.
You are talking about scientific bias here. Yes you can set up an experiment to prove whatever you may be biased to get as a resultant. That is often the problem with so called experimental outcome if you have a per-existing notion of the outcome. Very hard to overcome without third party intervention that has no stake in the outcome.
And I don't believe it needs to be conscious ... if the experimenter has deeply held underlying beliefs then all sorts of unconscious signals are passed through to the subject, with body language and tone of voice; and the strictness with which various procedures will be followed, the level of "fussiness" will vary depending on the thinking of the experimenter ...
Frank
I love the belief that scientists have that they have full control over the impartiality of "controlled" tests. I would like do an experiment, of such "controlled" tests, where the subjects are purely random, but the people who are running the experiments are carefully selected; at a higher level, they are the "true" subjects. And of course one group of experimenters hold strong, personal beliefs about what the outcome "should" be, in one direction, and the other group, contrary.
What odds will you give that the results of these "scientific" tests will match ...
Frank
never heard of independent replication? - to be fully accepted controversial results, experiments supporting new claims have to be replicated independently - and the predictions of new theories tested with differing experiments when possible
a postdoc of my acquaintance said that a honest, professional "enemy" was a valuable asset to a Scientist's career - no one else would dive so deeply into your work, try to find flaws, give you incentive to work harder, force you to refine your reasoning
I'm talking that in some cases a true scientific experiment can not be performed. The case in point unless one produces a speaker for which removing high frequency drive energy does that and nothing else all bets are off. That is a speaker with no IM or FM under the test conditions.
Do the third octave noise test on an LP set up if you want to see how bad the IM distortion is for simple transducers that are accepted as audiophile quality.
Last edited:
It is known that most young humans cannot hear single tones of above about 20 KHz, yet there is evidence that people hear tones above 20 KHz when combined with other tones in the audio band. When there are two or more tones combined, the data about the various frequencies and their relative amplitudes is expressed in a complex waveform. My assumption is that the human hearing mechanism detects tones of above 20 KHz due to the complex waveform containing the data of those higher frequencies.
Whatever any individual discerns by hearing may well serve that particular individual, yet, it cannot serve as any prove to others.
This is why I rely on my ears when choosing pieces of gear for my audio setup, but I don't rely on others' recommendations (nor on published measurements).
Last edited:
Thanks for the pointer, Scott.
I well may repeat my tests with a seperate super tweeter that is switched on and off with the signal band-split around 25kHz or so with steep slopes but moderate ringing (eg linear phase Neville-Thiele filters). BTW, the tweeters are AMT's running in current drive (in a new commercial monitor speaker I had the pleasure to design) and at low levels their distortion is really rather low (up to the freqs than I can measure now -- planning for 100kHz Gefell mic to increase that range). For that question, since there are no perfect transducers and will never be, it wouldn't make sense to conduct these experiments at all?
For the mic and recording chain distortion, that is before brickwalling and therefore part of the musical instrument under test, whereas reproducing chain errors (false IMD products) are not and thus can be a problem.
That´s what i meant with ´inconsistent´ results of later research.
It was an argument in the discussion about Oohashi´s results that the usage of a seperated super tweeter wasn´t a good idea. And that is surely a point to consider.
Hamasaki et al. tried to shed light on it :
"Perceptual Discrimination of Very High Frequency Components in Musical Sound Recorded with a Newly Developed Wide Frequency Range Microphone"
Presented at the 117th Convention 2004 October 28–31
http://radio.feld.cvut.cz/~vlk/micp/kaviti.pdf
Last edited:
I also think to know why. No hf is more pleasant than bad hf content.
Guilty!
I've level-matched my latest active speakers with ears only and my listeners said they were ruler-flat
My goal is to have the most-faithful reproduction of the source, no matter what my ears tell me and whether I like the sound of it or not, period.
I believe that the probability is very low that my ears can be trusted more than measurements, as far as measurements go. And I generally "have to" assume that applies to everyone else's ears, too.
Believing in only your ears might be acceptable, if you are the only one who will ever listen to the systems you design or adjust, and you never discuss sound quality with anyone else. The exception MIGHT be someone whose ears have been proven, over time, by measurments of their results, or even by very broad concensus. So my statement probably should have been "BLINDLY beleiving in only your ears...".
The main problem with my statements above is probably "as far as measurements go". i.e. Are there even sufficient measurements available? Are measurements simply not expedient-enough to be practical? Are ears often "good enough" and so much more expedient that they usually "have to" be the preferred method? If so, what might be compromised and by how much? IS the result usually "good-enough", or not? And how would we (in general) know that?
I believe that the probability is very low that my ears can be trusted more than measurements, as far as measurements go. And I generally "have to" assume that applies to everyone else's ears, too.
Believing in only your ears might be acceptable, if you are the only one who will ever listen to the systems you design or adjust, and you never discuss sound quality with anyone else. The exception MIGHT be someone whose ears have been proven, over time, by measurments of their results, or even by very broad concensus. So my statement probably should have been "BLINDLY beleiving in only your ears...".
The main problem with my statements above is probably "as far as measurements go". i.e. Are there even sufficient measurements available? Are measurements simply not expedient-enough to be practical? Are ears often "good enough" and so much more expedient that they usually "have to" be the preferred method? If so, what might be compromised and by how much? IS the result usually "good-enough", or not? And how would we (in general) know that?
Gootee,
I agree that we as designers need to use both our ears and our instrumentation to analyze any sound system to have a complete picture of what is happening. Though we talk about the Golden Ears hear I would like to know how often these same people have that Golden hearing checked. Losses in hearing are often very slow to occur. So many think that this is only in the upper frequencies but I have seen this to be not the true picture to many times. I have had more than one friend who has had their hearing losses in the mid-range vocal range rather than just the upper frequency range. Am I to believe that what they are hearing is going to be the same thing that I am hearing, I think not. Working with loudspeakers we also have to look at more than just frequency response, we also have to look at polar responses and that I think is just asking to much of just your ears to be accurate. Phase response is another issue as is any decay problems that may be present in a speaker. Do you really think that you can accurately identify a frequency with a long decay time in a dynamic music passage? With tones and measurements we can identify the frequency or frequencies accurately and this we can use to identify what the problem is and how we are going to solve that issue. So far I am only talking about loudspeakers and assuming a perfect electronic reproduction chain. So I have to agree that we have to use all of our senses along with electrical testing to have an accurate picture of what we are looking for.
I agree that we as designers need to use both our ears and our instrumentation to analyze any sound system to have a complete picture of what is happening. Though we talk about the Golden Ears hear I would like to know how often these same people have that Golden hearing checked. Losses in hearing are often very slow to occur. So many think that this is only in the upper frequencies but I have seen this to be not the true picture to many times. I have had more than one friend who has had their hearing losses in the mid-range vocal range rather than just the upper frequency range. Am I to believe that what they are hearing is going to be the same thing that I am hearing, I think not. Working with loudspeakers we also have to look at more than just frequency response, we also have to look at polar responses and that I think is just asking to much of just your ears to be accurate. Phase response is another issue as is any decay problems that may be present in a speaker. Do you really think that you can accurately identify a frequency with a long decay time in a dynamic music passage? With tones and measurements we can identify the frequency or frequencies accurately and this we can use to identify what the problem is and how we are going to solve that issue. So far I am only talking about loudspeakers and assuming a perfect electronic reproduction chain. So I have to agree that we have to use all of our senses along with electrical testing to have an accurate picture of what we are looking for.
Thanks for this, my main concern is the design of the experiments. I didn't read every word but the above paper is interesting. The biwa is a good choice of instrument (the sibilance of a shakuhachi played by a master would also be interesting). One observation the mic has a large rising response past 20kHz and beyond that ripples that look a lot like those of a high order filter. With a strongly plucked instrument like the biwa, I would think it important to address the effects on the peak and shape of transients. I certainly am willing to entertain the possibility that the leading edge of transients can shift in localization. I definately have heard the sibilant sounds on instruments have an off image "noise" component added. OTOH I have perfectly enjoyable CD's and LP's of traditional Japanese instruments.
BTW - These instuments make sense as being right there in the room with you (as does solo acoustic guitar), I find almost any recorded examples to be so far from the reality I would probably look for something more fundamental wrong.
Last edited:
Candidates for loudspeaker evaluation at my former client Harman's facility undergo hearing tests beforehand. The little secrets that emerge: most professional audio reviewers have abnormal/deficient hearing, and their evaluations are usually less internally consistent than even those of untrained listeners. The consistency of the responses from trained listeners is remarkably good.
Ricardo
Thank you for the info.
Do you remember the frequencies and slopes of the band limits and where the filtering was applied in the reproduction chain?
Brad
This is in full agreement with Ricardo’s concluding memories.
George
A problem with that is you are usually testing their hearing amplitude response rather then their listening abilities. Listening includes the mind's interpretation of what is being heard.
I have tested my own hearing to 45k. I used a speaker able to go to 100k. A 15k test tone with 3rd harmonics on and off, verified with a 1/4 inch microphone and a FFT analyzer.
- Status
- Not open for further replies.