My ears are very very sensitive to distortions but I like some of the hard metal musics. I don't think that it is the distortion that is loved. I like the "spirit", the "adrenaline", or simply the typical musicality of such music genre. The distortion is the trade-off for the musicality...
I think "harmonic distortion" can be misleading. There are many different kinds of distortions, which are perceived in many different ways too...
There are reasons why distortion added to electric guitar sound is hard to perceive. For ease of identification, our brain always need two things to be compared and one of it must have special characteristics such as "painful", "sound wrong", etc. "Just different" is hard to perceive (for example comparing 1kHz to 1k2Hz).
Probably correct, but probably wrong too. Unpleasantness is something hard to perceive. Easier to perceive is when that distortion modifies the sound (usually frequency response) such that it appears "wrong" to your ears. And of course, you are more familiar with the guitar sound than those young people so you know better when the sound is "wrong" or not.
To test only for unpleasantness, the sound must be a simple tone (pure tone plus one of its harmonics). This will be hard, and because you tend to think that all opamps sound the same, we readily know what to expect from such test.
I didn't know if it is more preferred today than it was in the past. If so, may be this is also caused by the better speaker in our world today. Most old speakers have squawker that has terrible distortion. Now we have much better midrange/fullrange/tweeter...
Hehe. Characteristics of people who aren't sure of what they are saying is often they put that "exit clausul" (I don't know the proper term for this) such as this: "But used properly, in a typical audio application, I’ll challenge anyone to a listening test and bet they won’t be able to tell the above three op amps apart. The only catch is it would be a blind test and the listener won’t know which op amp is which"...
"Used properly" is the first clause (What a qualitative statement!). This makes their statement can NEVER be wrong...
"It must be a blind test" is the second clause. This makes the chance for being proven wrong becomes much smaller...
As far as testing for unpleasantness with distorted guitar tones, I find that more dissonant intervals can tolerate lesser amounts of distortion before becoming excessively unpleasant. Perfect 4ths and 5ths can sound pleasant enough with quite heavy distortion. 3rds, less so. Maybe diminished 5ths, the least.
Therefore, I like to adjust distortion levels to be not excessive for a particular riff or passage, depending on the musical harmonies involved.
But we must take that Ancient Alien thing very seriously!
It is after all scholarly based peer reviewed scientific journal formatted for
modern television audiences.
Back to our regularly scheduled DIYers.
Cheers,
Sync
It is after all scholarly based peer reviewed scientific journal formatted for
modern television audiences.
Back to our regularly scheduled DIYers.
Cheers,
Sync
Just about everyone could readily agree upon what "used properly" means. That isn't an out.
And why do you think a blind test is invalid? If you are given a sealed box that takes +/-15V, an input and an output, and you cannot tell me if there's an opamp or just a pass through wire in there, then that is rather definitive about the "sound" of an opamp, is it not?
It may not be that simple. What if 96 people out of 100 can't tell, but the other 4 can tell 68% of the time (measured over 100 trials)? Does that mean a difference is humanly hearable or not? Maybe it means it is barely hearable, but only by rare individuals? If not deemed hearable, at what statistical level should something qualify as humanly hearable or not?
As to the issue of blind testing, I hope we can find a way around that as being an issue. Maybe training blind people to recognize opamp differences would serve to convince skeptics there is something a bit more complex going on than might otherwise be expected. As far as I know, nobody has ever tried it, but I wouldn't be surprised if some blind folks could learn how to do it with statistically better than random results for many cases.
Consider this: have you ever looked at a spot on a wall, or a place in the nap of a rug where you can just make out a crude image of a face? Not a real face, but something like a crude sketch of one. If you look away and look back it can be really hard to find again. But if you look at it for awhile and try to remember it, you may be able to find it more easily next time. Especially if you know where to look. If someone shows you lots of sections of walls or lots of rug surface, finding the exact spot and looking at it the same way again might be impossible. But if you had a good clue where to look, you might be able to find it. It other words, if you are allowed to peek. But once you do find it, you might be able to show it to other people and if they look at it from the right angle with the right light they may see it too and agree there really is something there, a kind of pattern, that looks like a crudely rendered face.
Why can't the same kind of thing happen with sound? Once you learn how to listen to a particular amp, maybe you can find the little feature of distortion you found before. It's still there, and you could probably talk other people into listening the right way so they can hear it too.
This is all a kind of learning, where part of what has been learned is an association of what to look for or listen for, depending on some hint(s) or priming to get you close enough before the final pattern recognition can take place.
Of course, if this kind of thing can happen at all, there are probably a lot of more obvious cases where more and more people can see or hear whatever it is a lot more easily.
This is not to say that all perceived differences are real. They aren't. But figuring out what is real and what is imagined is not trivially simple. It can be rather complicated, but I think we could probably start making more progress on it, if someone or some group is motivated enough to start doing the work.
Last edited:
For ABX testing, you will hit 50% if you just flip a coin and base your response on that. Typically, in ABX testing, you need to see 90 or 95% correctly to assign any statistical significance to the result.
Next, consider that in ABX testing, audiophiles are pretty poor at detecting the difference between 24-bit content and 24-bit content that has been rendered at 16-bits. In other words, people that really care about this stuff generally cannot hear a difference that is readily detected by cheap audio equipment.
Then, understand that a good opamp outperforms the most expensive test equipment on the market today. That's right, a $3 opamp dramatically out performs a $10,000 piece of test equipment. For this reason, a bunch of techniques to intentionally degrade the performance of the opamp is needed just so the test equipment could measure it.
So, when you say you can "hear" an opamp, what you are really saying is that you'd score 100% on the 16 versus 24-bit test, and then you'd be able to hear 2-3 bits beyond the 24-bit test.
i'd suggest you take any number of the 16 versus 24 bit tests available on the web today and see how you do on those. If you hit 100% score on those tests, then you MIGHT have a fighting chance at hearing an opamp. but just know the opamp is far, far better than a modern 24-bit source material. Modern 24-bit source material is actually about 20-bits in terms of noise and distortion. A modern opamp is about 22-bits in terms of noise and distortion.
No, it is a well-defined engineering term. There's nothing remotely misleading about it.
You are exactly repeating my point. To me, heavy harmonic distortion sounds painful and wrong; to many younger people, it doesn't, because they've spent years listening to heavily distorted guitars in music.
A lot of quite ordinary people (i.e., not trained musicians or people with perfect pitch) will notice if you play back their favourite song pitch-shifted up or down by a single semitone. That's only about a 6% change in frequency.
Your example is a 20% change in pitch. It's huge, enormous, gigantic. Unless they're played to the subject days apart, nobody with normal hearing will have any trouble telling them apart.
Didn't you just say that it is easy to perceive something that is "painful" or "sounds wrong"? This seems to directly contradict your earlier statement? 😕
I don't "tend to think" - I know so. It's very, very simple to prove it: make an opamp gain stage, divide the output back down to the same size as the input, subtract one from the other, and measure the difference signal.
With any decent opamp, used properly (I will explain those two words to you later, since you didn't understand them), you'll find that this difference signal is too small to be heard by human ears, and usually buried in the noise floor itself.
For example, if the distortion is 0.001%, the error signal is a hundred thousand times smaller than the actual signal being amplified; that is to say, any nasty noises the opamp is making are at -100 dB compared to the signal. Nobody can hear distortion that's 100 dB down from the signal, buried in the noise floor.
Most people today listen on truly lousy-sounding little earbuds, and never go anywhere near a good speaker system.
But if they did have a good speaker system, it would only make distortion sound even harsher. What happens if you add a good tweeter to a heavily distorted guitar amplifier? The sound becomes so harsh that it is intolerable. Instead, guitar speakers are designed to have a low-pass characteristic, filtering out the harshest-sounding harmonics (almost everything above 5 kHz or so.)
Music has been getting gradually more discordant for centuries. When electronic music arrived, intentional electronically generated harmonic distortion became part of it, and the amount of distortion that people will accept (without putting their fingers in their ears and running away) has increased dramatically between, say, 1930 and today. Clearly, there has been a long-standing cultural trend towards incorporating higher and higher levels of formerly unpleasant sounds into music.
I would say the prime characteristic of people who don't know what they're talking about, is that they say things that are nonsensical, imaginary, and impossible to demonstrate objectively. 😀
Sorry, you completely failed to understand me. I'm talking about engineering facts, namely, every op amp has limits, and if you use the op amp in a poorly designed or inappropriate circuit, you will find those limits.
For example, different op amps have different equivalent input noise voltages; a TLO72 is a few decibels noisier than an NE5532, for instance. Used at the intended signal levels, both are sufficiently quiet that you won't hear any noise. But if someone sets them both up with a voltage gain of 10,000, and feeds them a 10 uV input signal, you will hear lots of hiss from both, and you may hear more hiss from the TLO72.
This only proves two things: one, that person doesn't have the faintest clue how to use an op amp; two, the 5532 has slightly less noise. (Which isn't a surprise, since it's right there in the data sheet.)
There are lots of ways in which to use an op amp improperly; you cannot get any meaningful results from those cases. It would be like saying that Honda's are better than Hyundai's because licking ice-cream off a Honda tastes better than licking ice-cream off a Hyundai...
You mean, it makes the chances of the listener fooling himself/herself into thinking he/she is hearing something, much smaller! 😀
This is very solidly proven, and there are decades of experience, and thousands of test cases to demonstrate it. If you can see the amp or speaker, you will fool yourself into hearing differences that don't actually exist. It's just the nature of the human brain.
If, on the other hand, those differences exist, and are sufficient for your senses to detect, there should be no problem at all hearing them in a blind test!
-Gnobuddy
Because, imho, blind test only show human weakness in using their short term memory and listening skill. It's like a blind man and a wall. Just because the blind man doesn't have the ability to see the wall it doesn't mean that he can pass through it. This is an opinion from a man who can see the wall...
If only I can be a representative of human in proving that most opamps sound different in typical circuit through ABX, I think the case will be closed.
From your statement, I think you have made wrong assumptions...
You only think about equivalence in distortion, which is wrong. Opamps have super low distortions but what we hear is distortion from transducers/speakers. Opamps case is very similar like (power) amplifiers case. Do all amplifiers sound the same, especially when the distortion is super low like opamps?
What we hear is the sound coming from the speaker. So we shouldn't only care with amplifier "internal" performance but it is necessary that the amplifier does not trigger uncontrolled behavior of the speaker...
Simple trade-off in amplifier design is distortion versus speaker damping. This is very easy to test and very easy to hear (for me, but you can try it). Many amp builders here do not understand this, which is easily seen when their amplifiers have ridiculously low distortion (for relatively simple circuits)...
This damping issue also applies with opamps. Let's assume that the load impedance for an opamp circuit is 47k (amplifiers' common input impedance). Do you think all opamps will have equal capability to drive this load? What if the opamp is inverting, will that worsen the case? (answers: "no" then "yes").
Where is the test? Isn't there anyone hit 100% already?
I was talking about your perception. When you think you hear a "distortion", there is no guarantee that you know exactly what engineering variable it is, especially when we hear music and not a pure tone.
No. I was questioning two of your conclusions: (1) Youngsters do not feel heavy HD painful because they have spent years listening to heavily distorted guitars. (2) You can detect the extra HD better than the youngsters because you can hear the unpleasant effect of the HD.
My prediction is, if both of us were given a blind test regarding HD unpleasantness with simple tone, I believe that I would perform better. But if regarding sound of guitar, you may probably perform better. The point is, it is not about your ability to perceive unpleasantness in HD, but your familiarity with guitar sound. (This is how our brain works).
May be I wasn't clear. Try to listen to 2 tones and then guess which one is higher in frequency. You will be surprised. It's like why only very few people can precisely describe a note if it is played without reference.
Yes, the brain wants to compare with something in long-term memory. No contradiction, because I supposed to mean "unpleasantness of HD", which is your "topic".
Read my earlier post regarding load impedance.
Earbuds are better than old squawkers.
No, distorted sound may not only come from the source, but from cone breakups. Heavy metal sounds good through good speakers.
If there is no signal above 5kHz, then why should a good tweeter produces signals above 5kHz? Ime, most unpleasant signals, such as coming at high SPL with guitar music, centered around 1kHz. This is squawker (midrange) territory.
Hehe 😀. It is always nonsensical for those who don't understand. Of course, it is possible to demonstrate. But there is no point to demonstrate something that "nonsensical" to those that doesn't have their mind ready to change.
Don't forget that when people say that opamps sound different, they are "different" in certain typical applications. It is your task to provide a circuit where any opamp will sound similar because they are used within their limit. Of course it must be a usable circuit (such as a preamp). That is why it is mentioned "typical audio application". For example, an opamp-based preamp which accept signal from a DAC and output to an amplifier...
Isn't the issue with circuit design is that all transistors and ICs have limits that prohibit them to function as a useful device? (Take for example the small output swing of the opamps)...
BTW, in my DIY activity, I'm not comparing opamps with the same circuit, because I know each opamps need different treatment. So you are free to post 2 different circuits with 2 different opamps where they are both used properly.
True. I have had no problem...
If you can't do it based on short term comparison, your chances of long-term are even worse. Humans absolutely suck at long term audio memories.
The power amp stage distortion is 100X higher than line-level opamps. Therefore, if you cannot ABX a full lineup (buffer, volt gain, current gain etc), then by definition the opamp is transparent.
47K isn't a challenge at all for a modern high performance opamp. At all. Modern opamps are driving into headphones (16 ohms) with killer distortion levels rivaling what $10,000 audio equipment can measure
Not even close. Read the 3 part series linked below and you'll see that everyone fails this test. In spades. Audio engineers. Pro musicians. Those listening on expensive system. With speakers or headphones.
Everyone has about the same accuracy, and performs about the same as if a coin had been flipped.
Archimago's Musings: 24-Bit vs. 16-Bit Audio Test - Part I: PROCEDURE
It looks like the Hardware Reviewers and Musicians preferred the 16-bit by a significant margin. That is not surprising. Some 24-bit sounds worse than 16-bit. That's what I found when I went to an online store that sells 24-bit audio recordings. They all sounded the same in some way, and it was ugly.
Separate out the information from the the rest of the signal (spatial cues, exact frequency, timbre etc etc).
We indeed remember the information part fairly well - that's the bit that for the most part ends up in LTM - but the rest not. It gets ditched.
I suspect that with training you can improve it, but we are not designed for it naturally.
You've lost me. To me the information in the voice is what's being said, however I'm talking about recognizing which person is speaking.
Perhaps we're at crossed purposes - to me seems perfectly natural that people recognize familiar voices.
There is misunderstanding regarding the terminology of "long term memory". What I mean with "long term memory" is part of information that is stored in our brain for long time (such as voice of people we know)...
Only by comparing a sound to another reference stored as long term information like that we can easily hear the difference. This is a known trick to memorize things and let it stay longer in the brain. We have to give some "meaning" to it. For example, when we want to memorize that "DIYA was born at 12/6/2017", we can link the date to something already in the special location in memory such as "DIYA was born exactly one week after my beloved dog was born". Principally this is also similar to what is called "mnemonic" and "mind mapping"...
The trick with blind test, we need to have that reference in our long term memory. If nothing there, we have to create it. Listen first (un-blind) until you can recognize something in the sound.
I have something to say regarding prolonged listening but I better not say anything right now.
Sufficiently transparent, may be. Remember my statement that human skill is so poor in the ABX, but it doesn't mean that everything sound the same... But if my listening skill is used as standard, I don't mind 😀
Again, you only focus on distortion. Even distortion is only a single parameter. What is the fundamental frequency? What is the level of the n-th harmonic?
Okay, thanks, I will look into it. But remember my statement: it is a different thing than the opamp issue.
I did a quick read and I think you made a wrong conclusion. There exist people who could correctly identify the 24-bit over 16-bit!
The point made by the author is that they are just a few, equal to the possibility that it is just "by chance". This is the author inability to understand and to properly use Statistics!
There is nothing new either with the fact that human are so poor in doing this audio thing. People are just reinventing the wheel. Why not just study and understand what have been studied by previous experts... 🙄
I'm speechless...
50% means something is completely random, assuming no other issues with statistics or testing methodology. 90 or 95% is completely arbitrary as a standard. 95% confidence levels are often used as a de facto standard for confidence levels in medical research, and human hearing ability studies should probably be classified as medical research. However, the 95% refers to the chances of a study being wrong out of statistical chance alone. It is not a figure used as a standard for defining statistically significant effects levels. Those numbers are typically much lower.
With a proper reproduction system, listened to in the near field, free of reflections, the differences between 16-bit, 24-bit, and high quality compression schemes are all quite audible. That isn't to say that higher resolution always sounds better. It doesn't always, and depends a lot on source material and ADCs used.
You are making a rhetorical argument for debating purposes here, rather than trying to seek out scientific truth. What people can or can't hear is not a function of the price of test equipment.
I am not really saying that and did not say it. I generally have no trouble hearing differences between 16-bit and 24-bit audio on my system, and in the near field. I also don't think a single opamp necessarily has a distinctly identifiable in circuits that are the least challenging to opamp capabilities. Sometimes we need to use circuits that where some differences are discernible. Also, again you are using debate tactics by speaking in absolutes. Science can never prove anything conclusively, it can only continue to reduce doubt as evidence mounts, and even then theories collapse with only one real counter example. Overconfidence bias is a big problem of human rationality, we all have it, and some may have a more or less than others. It seems like you may have ended up with at least your fair share. https://en.wikipedia.org/wiki/Overconfidence_effect https://en.wikipedia.org/wiki/List_of_cognitive_biases
I did look at the results of the one test you referred someone else to. I don't agree with your assessment of the results. And even at best, there are a number of unanswered questions and issues the author of the article should address.
- Status
- Not open for further replies.