The pictures keep getting posted off and on of vibrating disk modes. Now think about how some of the most popular recording mics have large disk diaphragms either edge or edge and center constrained where the primary resonance is well down into the most sensitive audible region. It is very easy to at least hypothesize that mics with sibilant reputations have something going on.
It probably is, but not stationary. Not something easy to see as a spectral peak on an analyzer or as some kind of shaped noise. Because music is constantly changing, and the distortion mechanisms are multiple, small, and very dynamically changing, it doesn't follow the model of a simple, single distorter.
We are getting better at measuring signals buried in noise, such as by fft techniques, but the brain does that too. Brains can extract signal below noise. For most people, the brain is best at that for recognizing threatening language buried in noise. That's probably because there is some survival advantage to it and natural selection prefers traits and abilities of survivors. There may have been research to see how good average people are at detecting tones buried in noise, but not someone saying, "There's jcx, get him! Hold him for me!" from behind in a noisy restaurant. In fact, that's probably why schizophrenics hear threatening voices, a little too much dopamine and false positives from the System 1 threatening language detector is the first thing to start malfunctioning. It then pops in to conscious awareness the recognition of a voice.
But, I don't think Bear has too much dopamine or is hearing voices. I suspect his brain happens to be organized in a way that is very good at detecting distortion buried in noise. What System 1 processing then pops into conscious awareness is something that may be perceived as smeared or sibilant (in the case of bear, at least, for sibilant), or whatever. The relationship between the experience of perception in conscious awareness, and the mechanism by which System 1 processing converts vibrations detected in the ear into such perceptions is poorly understood. The idea that what we hear and what we see is true reality is largely illusory. But it is an overwhelmingly compelling illusion. Thus, trying to translate bear's experience in conscious awareness directly to something in the frequency domain is probably a grossly oversimplified model of what is actually happening.
Last edited:
The JC-2 phono amp ran on ±13V power supplies and dropped about 1.5 volts across the output stage's source follower degeneration resistors. So it clipped well before an IC opamp, especially an IC opamp run from ±17V supplies as some books recommend.
Or something like the lme49860 off 20-21 volt rails. Or a discrete jfet + opamp hybrid if you want the advantages it provides. They're tools! 🙂
I figure if there was a corollary thread "What is right with discrete" there would be equally random discussion.
I attribute no causal factor for sibilence. I really do not know where it "lives".
Please go ahead and explain how and why it occurs, that is the objectionable case per the description that follows.
What I said is that while the natural voice has what is called "sibilence" that it is not the same as what many electronic reproductions create. The primary difference is that humans hear the naturally occurring sibilence of a real voice producing it as (in almost all cases I know of) unobjectionable and "in concert" with the rest of the vocal sounds, or "natural sounding". Whereas the sibilence produced or created or reproduced (improperly) by electronic means (of the type being discussed) is heard as objectionable, at odds with the rest of the vocal sounds and "out of proper place", and so "unnatural sounding".
The most simplistic language used contrasts the two: "relaxed" vs. "strident"?
Last edited:
Bear, some proper sense in this discussion, necessarily long winded but accurate never the less.
Dan.
The two examples need to be fabricated in exactly the same location/orientation/conditions and exactly the same point in time, in other words it is actually impossible for two examples to sound exactly the same.
Dan.
Something is <wrong> with opamps,
Simple balanced out of dac with LM4562 sounds better than many tube and super expensive dacs, all can be assembled for 100$ including box, dac, and power supply! and can drive balanced headphones straight from the output!
Simple balanced out of dac with LM4562 sounds better than many tube and super expensive dacs, all can be assembled for 100$ including box, dac, and power supply! and can drive balanced headphones straight from the output!
More great input, and 'false' sibilance is one of my first criterion when assessing reproduction/reinforcement.
It's but one product of chaotic behaviour, sorry no more clues.
Dan.
Perhaps. But we can be sure that an element of that must be teaching them some facts, as without facts people have nothing to think with. Facts are not sufficient, but they are surely necessary. Modern people seem to be resistant to learning facts; they prefer to learn standard answers to standard questions. They believe this is the purpose of education.Markw4 said:
Many years ago, someone in my class in school asked the maths teacher "Miss, why do we have to learn maths?" Her reply was "Learning maths teaches you how to think". That was true then. It may be less true today, because UK schools now teach less maths than they did in my day and the pupils are expected to think less - questions now tell them how to find the answer, but when I was in school we had to decide for ourselves which of several possible methods to try (e.g. to do an integration, or to do a geometric proof).
You can ask questions like "what bandwidth do we need?". Is 20-20kHz sufficient? At what point in widening bandwidth do we hear no further improvement in reproduction?Markw4 said:
You have completely missed the point. The question is not "do I like it?" - who cares whether you like it or not? The question is "does it sound like a violin?". Can I tell whether it is a violin or the sound of a violin being reproduced; how sure am I which is which?
On the contrary, this is doing science to answer a specific set of questions. What amplifier performance is needed for high quality sound reproduction (not communication, not PA, and definitely not instrument amplification). It has produced useful numbers. The question is whether these numbers arrived at over 50 years ago need refining. Some people (without much in the way of evidence) assert loudly that the numbers are so useless that we can ignore them - but then we find that they are not usually interested in sound reproduction because they keep talking about what they like (which often involves limited bandwidth and highish distortion). Meanwhile, someone else seems to assume that all we need to do is achieve 10x or 100x better than the numbers and then we need not worry at all; perhaps forgetting that all engineering is a compromise so if one number is 100x better than needed then some other number (which we might not yet know much about) could be 10x worse than needed.
No! The next level is "does it sound like the real thing?". That is all. For hi-fi there is no further level. Audiophiles (i.e. someone who listens to audio equipment) may have further requirements but people who listen to music will want to stop here.bear said:
Only non-linearity can change dynamic range. Filters are not supposed to be non-linear.
It is certainly old research, but why assume it is flawed? Ears-only tests seem to mainly confirm it, when they are carried out. It does not support belief in our own senses; on the contrary, it tells us not to believe them as they are easily fooled. It is the 'people who hear what you hear' who believe their senses, against all the scientific evidence.Markw4 said:
No. They keep repeating the same old arguments supported by the same old unsighted listening tests that what your senses tell you in sighted listening tests is probably not real.
Do you mean FR-issues? It is a bit more complicated. I know that energy storage issues in tweeters will cause sibilance. So it can be a non-LTI issue, not directly connected to the FR (although it will show up there as well. You just can't reliably EQ it out. And it will show up in the CSD). I think Scott hinted at this for microphones many posts ago.
Of course the sad part in this is that amplifiers (both pre and power) designed on this forum in some cases get to the point where distortion products are lower than the brownian motion of air molecules in the room.
Distortion products however are measured using what as a load?
So, the results may or may not be the same as said amplifier achieves in real use.
And I am reasonably certain that your position is that any two amplifiers of sufficiently low distortion as measured by the usual means are indistinguishable. Which, again, may or may not be the reality.
Two points.
First, at what threshold do these "distortions" become audible?
You cited them, Scott cited them, so beyond waving a hand at them, how does one deal with them in practical terms?
(also what about cascaded increases in this type of distortion as one progresses along the signal path??)
Second, I agree that you can not EQ this sort of thing out. But, are you suggesting that such distortion is inevitable due to the limited performance of "tweeters"? If so, that implies that nobody has a system without objectionable levels of this??
I asked Martin Kantola designer for Pearl Microphones years ago (they have a unique rectangular diaphragm) and he thought it was a possibility and worth at least studying. He built the most amazing ambient field mic out of NOS Neumann capsules and incredibly rare VF14 tubes.
Yes, but then why can or does one perceive what is being heard as a change in apparent dynamics/compression, and a change in the perceived soundstage dimensions? In this example the "bits" remain unchanged at the source.
On learning: I am not against schooling, or classical education. Absolutely, there is merit to it. And I agree it is good to learn as much math as one's brain is able. But, the fact remains that IQ is very good predictor of ability to learn what is taught in school, and respond well to demands to produce what is taken to be evidence of good thinking. Once out of school, high IQ and good school performance have very little correlation with life success, good thinking in areas not represented by IQ, such as dealing with wisely with other humans, saving for retirement, healthy living, etc. In other words, there is no evidence that people learned to think better about most things and make wiser decisions for most things in life as a function memorization of facts in school.
The ability to do things like form arguments, produce lists of reasons in support of, or against, particular position, mostly existed before college, and developed as the brain matured through teen years. I don't know about you, but statistically, I would expect you were about as good at arguing as you have been here when you were in high school, or maybe junior high. You probably weren't taught it in school, but in college you were required to demonstrate you were good at it.
A few years ago, looking through some old papers, I found an essay I wrote in 7th grade. The writing and thinking demonstrated in it was remarkably close to what I can do today. In other words, when my brain developed enough, I could do something well enough to pass college English requirements, but they really didn't teach me anything later there that makes me think better today. They just made me demonstrate my best work more regularly.
Regarding my comments on the live verses memorex question, my response in that case was intended to be more philosophical. With regard to some of your objections, you haven't really answered the question I was trying to raise, "how do you use subjective listening tests to guide design in a scientific or methodological engineering way?" As a practical matter, I suppose you could take listener comments, such as from someone like bear, go back to the lab and see if those comments help you sort out what you need to do to make the amplifier better. Maybe they will give you some ideas for investigation. But if your equipment reveals nothing, you will just conclude bear is nuts, and useless. If your equipment does reveal some problem, then why did you need bear to get you to take some measurements you could have and probably should have taken anyway?
The comments I made on believing one's own perceptions come from modern research, as do a better understanding of in what cases humans are irrational in terms of believing things that are not scientifically true. I find no evidence according to that modern research that bear is irrational based the beliefs he has formed and stated here.
And, he has not been carefully tested in a way consistent with modern researching findings about hearing small differences in sounds to see what he actually can do. So we don't know, or we don't have that type of information available right now. But I suspect he is an outlier in terms of distortion hearing acuity. He is at one extreme end of a distribution. Past research has not focused in microscopically on that segment of the population, or learned how to accurately identify such people by measurement methods most sensitive for that purpose.
When I say that past research is probably flawed, I don't think it is an arbitrary assumption. I mean that it produced results that largely apply to average people in the middle of the statistical distribution. It would then be an assumption to believe that there are no long tails on the distribution. It hasn't been properly investigated. For the purposes of our discussions here, the research would then be flawed, incomplete, or misleading, for application to some of our interests here where we want to know more about the potential utility in the pursuit of better amplifier design from data based on reports from people that may have exceptional hearing acuity for distortion.
- Status
- Not open for further replies.