Sound Quality Vs. Measurements

Status
Not open for further replies.
Disabled Account
Joined 2012



A few people have perfect pitch, they can say "that is A#" or even hear a half note difference. Normal people need to hear a few pitches near each other to know it, like a scale.

That's volume, FR and pitch. Few humans have perfect pitch or perfect volume, perhaps since it's not very interesting or useful, I'm not sure.


Kthx for reading or any comments.



I have 'perfect' pitch. I have only found it useful to tell someone if their turn-table was running slightly fast or slow. Or if their instrument is out of tune. Perfect pitch IMO just means you can judge the tuning/tone without a reference to compare immediately at hand. The reference is held in memory.

From what I know and have experienced.... when I connect those dots, I discover that all hearing and its interpretation is learned. Including perfect pitch.

If you have not heard the sound of a tuning fork used to tune with.... you dont have a way to learn or develop 'perfect pitch' hearing.

The same with hearing distortion.... you learn what 2H and 3H etc sounds like by comparing the measured with listening until you have a memory for that. And, if you use speakers, such as the new Quads, you can detect with "perfect' clarity very low levels of distortion or added tones which should not be there. Even identify the harmonic. You then start the development of a memory for certain types of distortion and are 'tuned' into it.

Its all learned and anyone can learn it. They just have to get that kind of listening training/experience.


THx-RNMarsh
 
Last edited:
Toobs,

Makes sense as pre-amplifiers, always have one on hand and it makes the most sense for them really , linear devices , no output trannies, very good as a small signal device . Toob Amplifiers on the other hand bring nothing to the table over SS , their cloudy sluggish presentation always duly noted, there are some exceptions, especially when taming wayward speakers , for the most part , not...

:drink:
 
Last edited:
AX tech editor
Joined 2002
Paid Member
And none taken, Jan. It's really about two things:

1. I very clearly said that excessive NFB >>> may <<<, not "must" or "will", possibly cause TIM related phenomena, which is, I believe, quoting one of Otala's original conclusions from 1973, and

2. I don't think anyone here would argue that designing for wide open loop bandwidth and low global NFB is just the same as designing for 60+ dB of GNFB and damn the open loop bandwidth all the same. It is therefore reasonable that the end result will also be different, if only by a little.

All I am saying is that to me, that "just a little" seems to be the difference between whether I like it or not more often than not. This seems to be the inevitable conclusion of my 6:1 ratio of preference. It's not a question of right and wrong, it's simply a personal preference.

And it certainly doesn't make either Otala or Cordell right or wrong.

Yes these are good points, and it may well be that there is an audible difference as mentioned by you. But we should be careful to ascribe this to a single factor. For instance, if one would design a low or even zero nfb amp, necessarily the internal stages will be different than with a high nfb design in terms of stage gain, stage distortion, bandwidth etc. And slew rate is also a factor to consider. But concluding that in general a low nfb design sounds better because of lower TIM (and I know you haven't made that statement) is very much a prejudice to feed an existing belief.

Coincidentally, I am now working on the first Linear Audio book (not written by me!) which I hope to publish this fall, and which concerns itself with internal stage gain linearity and symmetry and internal error correction, as well as the trade-offs between the nfb factor and internal stage performance.
The designs presented range all the way to a zero nfb design with incredible performance. But there are many, many factors involved, and the trade offs are many, and that is why a whole book is needed to put everything in order and perspective.

It may be possible to state problems in a simple way, but the answers seldom are. As one punter quipped: For every complex problem there is a simple, easy to understand and wrong answer.

Jan
 
I do hope the book uses a method at least as systematic and illuminating as AES E-Library Estimates of Nonlinear Distortion in Feedback Amplifiers

its rare to see audio amp feedback loop design commentary up to the level of the 1963 http://www.amazon.com/Synthesis-Feedback-Systems-Isaac-Horowitz/dp/0123559502 understanding of feedback loop structural constraints

much less the insights of BJ Lurie's work - though you have to be pretty dedicated to follow his books

more of Self's approach probably won't be a big help - not that I dislike it - just don't see much future in that mode of exposition for advancing the art
 
Last edited:
AX tech editor
Joined 2002
Paid Member
Yes I know Cherry's paper. The book treats many factors and solutions in a very systematic way. When I am a little bit further I will post the ToC.

BTW The idea to assume a distortion free output and then see what kind of distorted input signal you need for that is not new.
Fred Waldhauer, a old colleague of Bob Cordell at Bell Labs (and the man who 'invented' the T-line) wrote a book in '82 or so with the simple title 'Feedback'.

He already had this idea that Cherry picked up on later. In simple terms, Waldhauer shows that his way of thinking makes the math to treat feedback much simpler. The feedback signal is undistorted, and it also solves the 'problem' that many people cannot get their head around what seems to be a correction after the fact.

Jan
 
I found the Artech "Feedback Maximization" heavy going but the examples are generally applicable to feedback systems - not specialized
the problems are that the nonlinear compensation analysis, justifications seem poorly motivated, I found it hard to see the "systematic approach"
I did find it for ~$15 with shipping though

the later Classical Feedback Control: With MATLAB® and Simulink book 1st edition is very buggy but much more readable - but still not exactly clear even if you can catch most of the equation and text errors
I haven't seen the 2nd edition - if it is just a clean up or a real improvement
 
Last edited:
valve amp that might be even 1% THD and is still hi fi with the right distortion spectrum
As Jean Hiraga pointed out the 5 th ( and 7th, 9 th) harmonic is the problem
Jean also said a high THD distortion curve reducing exponentially might sound less distorted than an amplifier that had only 5 tH harmonic
As speakers seldom go below 1% THD I am reluctant to think distortion THD is what we are listening to
As far as I know no one that has ever lived could hear THD below 0.1%.
|
I see you have presented the following points

- A high distortion number can sound like low distortion, it's necessary to......
-- analyze the spectrum
-- analyze the harmonic number, including the 5th, 7th and 9th
- A high distortion curve reducing exponentially might sound less distorted than an amplifier which only has the 5th harmonic.
- Speakers seldom go below 1% THD
- As far as you know, no human that has ever lived can hear below 0.1% THD

______

The first point. I've never heard of the 5th, 7th and 9th issues, I will look it up later.
Yes we need to analyze the spectrum, there are distortion spikes which we can not see with a single solitary number, such as "THD = X.XXXXX".
The spikes may cause some sort of harmonic or non-harmonic overtone, i.e. they have a particular spacing, which could translate into a tonal character.
The THD number "X.XXXXX" fails to reflect the masking envelope, i.e. "the masking effect of the signal itself".

An article on that here --- http://en.goldenears.net/index.php?mid=KB_Columns&category=264&document_srl=1335

Thus without reflecting the masking effect, the THD number does not correlate to human hearing, apparently.

Next point

Ok, I assume you mean a curve which starts high at 1 kHz and then decreases lower and lower at 2 kHz, 3 kHz, 4 kHz, 5 kHz may sound less distorted than a violent solitary distortion at 5 kHz.
I don't know if it will sound less distorted, but for sure, that should sound different.

Next

"Speakers seldom go below 1% THD"

Well......

1. Can we hear distortion within distortion? Like the cocktail party effect.

If you record twenty people talking incessantly at a party in a small room, can you separate the data you need of a single person in computer software?

I don't know if this is an accurate analogy or not, but apparently we need stereo for the cocktail party effect.
Therefore, perhaps in an efficient stereo system, we can 'divide the distortion' a little, by displacing it left and right in the soundscape.

In any case, I'm not aware of any evidence which indicates that transducer distortion masks amplifier distortion. Without that evidence, it's a theory.

2. I haven't looked into speaker distortion, but I assume some speakers have much lower than 1% THD.

Personally, I use speakers at times but I usually use IEM's, IEM's are my area. The data I've seen of IEM's shows pretty low distortion levels?

See example......

0.0766% - M.R.O.: Vsonic GR07 mkII
0.0427% - M.R.O.: The effect of headphone cables: SONY MDR-EX1000
0.0673% - M.R.O.: STAX SR-003 MK1: Electrostatic in the Ear

Next

"As far as I know, no human that has ever lived can hear below 0.1% THD"

I see two alternatives......

1. Audio enthusiasts playing with capacitors and op-amp's are all living inside a tricky illusion.
2. Audio enthusiasts playing with capacitors and op-amp's can reliably and vividly hear distortion under 0.05% THD.

I suppose the third alternative is......

3. Audio enthusiasts playing with capacitors and op-amp's can reliably and vividly hear something else, like PSRR, CMRR, EMIRR, time character, EMR, ESR, spirits, ghosts, witches, elves.
|
 
Last edited:
Yes these are good points, and it may well be that there is an audible difference as mentioned by you. But we should be careful to ascribe this to a single factor. For instance, if one would design a low or even zero nfb amp, necessarily the internal stages will be different than with a high nfb design in terms of stage gain, stage distortion, bandwidth etc. And slew rate is also a factor to consider. But concluding that in general a low nfb design sounds better because of lower TIM (and I know you haven't made that statement) is very much a prejudice to feed an existing belief.

Coincidentally, I am now working on the first Linear Audio book (not written by me!) which I hope to publish this fall, and which concerns itself with internal stage gain linearity and symmetry and internal error correction, as well as the trade-offs between the nfb factor and internal stage performance.
The designs presented range all the way to a zero nfb design with incredible performance. But there are many, many factors involved, and the trade offs are many, and that is why a whole book is needed to put everything in order and perspective.

It may be possible to state problems in a simple way, but the answers seldom are. As one punter quipped: For every complex problem there is a simple, easy to understand and wrong answer.

Jan

Well, using the open loop wide bandwidth approach actually involves a lot more than one approach to get it right, as I know that you know. One can't just twiddle the values of a couple of resistors and get it right, it's a proper appraoch, which implies it as the underlying idea of the whole amp.

Also, I am sure you've noticed, such amps tend to generally fall into at least medium speed territory in terms of slew rates, and often into the fast lane.

As ever, it's not quite so simple as when just chatting about it. For example, I haven't heard many zero global NFB amps, but those that I did never really quite sounded right to me, it's as if they were not quite in complete focus. So I do like to apply some GNFB, I find it's that last bit needed to get it just right, although I have not made very many amps, just 7 over the years. That's not counting those I did according to other people's schematics, like the Otala/Lohstroh (which I was an idiot to sell), the German made LAS Mega 1, etc.

Every design, no matter what approach t uses, has its own challenges and dilemmas, and none is really easy, all take hard work.

As for prejudice, Jan, I remind you of my statistics: 6:1 in favor of low NFB, wide bandwidth. If that is prejudice, then I am prejudiced, but I made my purchasing decisions governed by my ears, not my views on technology. Call it what you like, but that's how I hear things, and with a statistic like that, I'd say that was pretty conclusive. Of course, that's just me, and I certainly do not claim that this is the only right way to go.

Regarding your book, I do hope I will be able to buy one, and I do hope you will remember to let me know it's come out of the print whenever that might be.
 
I have 'perfect' pitch. I have only found it useful to tell someone if their turn-table was running slightly fast or slow. Or if their instrument is out of tune. Perfect pitch IMO just means you can judge the tuning/tone without a reference to compare immediately at hand. The reference is held in memory.

From what I know and have experienced.... when I connect those dots, I discover that all hearing and its interpretation is learned. Including perfect pitch.

If you have not heard the sound of a tuning fork used to tune with.... you dont have a way to learn or develop 'perfect pitch' hearing.

The same with hearing distortion.... you learn what 2H and 3H etc sounds like by comparing the measured with listening until you have a memory for that. And, if you use speakers, such as the new Quads, you can detect with "perfect' clarity very low levels of distortion or added tones which should not be there. Even identify the harmonic. You then start the development of a memory for certain types of distortion and are 'tuned' into it.

Its all learned and anyone can learn it. They just have to get that kind of listening training/experience.


THx-RNMarsh

Stanley Kelly found in circa 1947 that 97% of people have perfect relative pitch. If I remember correctly that was to listen to 1 kHz then set a generator afterwards. The pass mark was 0.1% or 1 Hz. On the other hand 80 % of people had hearing loss. Stanley reflected that it was a more industrial age then. The perfect relative pitch is the facility to learn a language. These test were done for the government. Mr Kelly turned down the tests a year or two before. A new condenser microphone made it possible to do some of the tests which were not possible before.

The voice is a non linear instrument. By exercising perfect pitch it is made linear. As far as I know to leap from perfect relative pitch to perfect pitch is a big one. That ability of 97% of people is no small thing and makes people discerning without words to say what and why. If I remember correctly PRP is the left brain or intellectual side and PP right brain, the artistic side (supposedly). When we speak foreign languages we are forced to use the opposite side of the brain to our mother tongue. That actually makes us diffident people in that language. Bilingual kids from birth are not like that. It is thought that they never truly master either language. It has no bad problems and is a worthwhile trade off. It is the esoteric vocabulary that suffers in number of words. That is very difficult for a Japaneses person as 20 000 written characters to degree level and 90 000 to PHD in some subjects. That is long before committing them to memory as pitch notation. The other problem is the characters were stolen from Chinese so don't fit perfectly. Eventually it seems likely they will give in and use western script.

IBM had very big difficulties building programs that could take voice dictation. Gaps that you and I call words are invisible to the computer. Weird but true.That is perfect relative pitch. Perfect pitch says that work is in C minor or similar.
 
Last edited:
Kastor. I think the tonality of musical instruments is not well enough known that 7 000 000 + people have any way of supporting the less than 0.1% THD criteria. The Ear has 30 % THD in exponential decay. It by excellent processing can ignore that. You can not know the sound you are listening to with that precision. Distortion is just timbre when not dreadful. To say more than that is an exercise is the absurd as it would be like me saying my eyes unaided can see individual bacteria ( I think there are ones that come close).

I do believe you might hear one amplifier sound different to another and by coincidence it is at the virus level of THD they differ. That is you are gifted as an electron microscope in your perception? I think not. It will be the ill defined TID or whatever is this weeks name for it. This is the Sun goes around the Earth observation. The poor man who proved against that took 7 years I think? The timbre of instrument is the Sun central argument. Very low THD is a harmless pursuit of people with no new philosophy. Mostly it does no harm. You bet when someone proves something worthwhile very low THD will be of zero importance. The new will be like knowing of ultra violet or infra red. Why didn't I see it ? Well, how could you?
 
Realistic simulation of a typical amplifier, which means nothing is assumed to be ideal, tells one a lot - most importantly, that the circuit does not behave well in transient situations, under load stress. Which of course is exactly what everyone hears, in the real world, with real amplifiers. So there are no mysteries, it can be shown why real amplifiers sound different - but while conventional measuring is relied upon to assess performance this will remain underground, ignored.
 
That is very difficult for a Japaneses person as 20 000 written characters to degree level and 90 000 to PHD in some subjects. The other problem is the characters were stolen from Chinese so don't fit perfectly. Eventually it seems likely they will give in and use western script.

IBM had very big difficulties building programs that could take voice dictation. Gaps that you and I call words are invisible to the computer. Weird but true.

Learning 20,000 characters may have some effect on the mind and give light to parallel skills. Perhaps it improves an ability to draw for example, the Japanese are pretty good at that.

It has been shown that musicians have much more grey matter. It has been shown that playing Tetris will change the shape of your brain in a few weeks.

It's quite nice to know that we have this neuroplasticity IMHO.

This says something about culture as well, it's not only learned behaviour, if you stay in one country for a long time you may in fact be rewired at the neuronal level.

Yes, it's difficult for a computer to interpret a single voice, let alone the cocktail party effect interpreting a voice while 20 are talking at once, just think about it, only one twentieth of that information needs to be analyzed.

Acoustic information like size and distance is not very well recorded and deciphered either.
 
Last edited:
Disabled Account
Joined 2012
We should put to rest the very old mark of 1% as being typical of todays speakers. Distortion well under 1% has been available for decades. If one is to hear at or below .1% it would be with a speaker of .1 or much less --- to be able to notice changes to tones caused by added harmonics.
Example :

66Quadfig10.jpg


Use a speaker like this to train your hearing acuity.


THx-RNmarsh
 
We should put to rest the very old mark of 1% as being typical of todays speakers. Distortion well under 1% has been available for decades. If one is to hear at or below .1% it would be with a speaker of .1 or much less --- to be able to notice changes to tones caused by added harmonics.
Example :

View attachment 429218

Use a speaker like this to train your hearing acuity.

THx-RNmarsh


Thank you for posting that Mister Marsh.

I suspected there's quite a few speakers with much lower than 1% THD.


The IEM's I linked to earlier, after looking a bit further I think they may actually be the lowest THD IEM's on his site.

I come from the portable audio realm which is why I didn't sign up at this forum for many years, very little portable audio here.

The portable STAX IEM is very nice, with the right fit. I can recommend it to anyone looking for in-ear experiences.

The Olasonic Nami / TH-F4N is a very high performance IEM as well.

http://www.aloaudio.com/earphones/olasonic-nami-82

It doesn't get very much better than these two, in my experience.

Just sharing some info.
 
This is what I REALLY said (italic added):

But concluding that in general a low nfb design sounds better because of lower TIM (and I know you haven't made that statement) is very much a prejudice to feed an existing belief.

It is, at best, an oversimplification, no better or worse than believeing that you resolve everything if you only use a lot of GFNB.

Call me a conspiracy theorist, but I believe things are never that simple and easy.

I believe it's all about achieving a good balance between the many factors you need to take into account. It always turns out to be something like that. Extremes never work for me.
 
You can not know the sound you are listening to with that precision.
To say more than that is an exercise is the absurd as it would be like me saying my eyes unaided can see individual bacteria //

I do believe you might hear one amplifier sound different to another and by coincidence it is at the virus level of THD they differ. That is you are gifted as an electron microscope in your perception? I think not.

Let me re-check this with you, so mean listening to THD under 0.1% is the equivalent of the naked human eye seeing bacteria, or let's say the human finger sensing at the nano-technology level?

Distortion is just timbre when not dreadful.

Yes, high distortion in vintage DAC chips is considered elite sound in some circles, such as the TDA1541, Lampizator loves that chip, right? It's most likely the distortion and harmonic or non-harmonic overtones, isn't it?

Like, a fetish for the way electronics can sound. The high distortion variety may at times sound weak, at times sound aesthetic or pretty, when executed in the right way.

That is you are gifted as an electron microscope in your perception? I think not. It will be the ill defined TID or whatever is this weeks name for it.

Ok, you believe me that I can hear op-amp differences and that this is due to TID or TIM.

I see, this is useful, since I've never heard of those terms until just now in this forum, when I saw scott wurcer say "it's a dead-end" and then DVV write his comments on it. From what I saw in this thread, John Curl supports it as well, Jcx not very much, I'm sure he has his factual reasons as well.
 
Last edited:
Status
Not open for further replies.