+1. A couple of books dealing with the nature of perception :
Auditory Scene Analysis: The Perceptual Organization of Sound by Albert S. Bregman
The Tree of Knowledge: The Biological Roots of Human Understanding by Humberto R. Maturana
Yeah, what the GedLee metric is showing is quite obvious: it is the integral of the product of
- Square of second derivative of transfer function, so a sharp kink in the transfer function gets more weight.
- And a weighting function that gives more weight to it depending on "where" it is in the voltage range. So it gives high weight to a distortion-generating transfer function kink if it is where the signal spends the most time (ie, near zero) but much lower weight at high output voltage, which is an area of the state space only visited on peaks. So when the signal doesn't go there, by definition, it doesn't matter.
These two signals contain the exact same error signal, except it happens at a different phase, ie a different point in the transfer function, one has crossover distortion near zero, one is clipping on peaks. THD spectrum is identical. Only phase of harmonics differ, but it is hard to interpret.
So they'd get very different GedLee metric, with the one clipping being less obnoxious, because it shows a transfer function that is well behaved around zero, whereas the other will distort no matter the amplitude.
The transfer function makes it obvious:
So, why not measure the transfer function? It makes the defect much more explicit, it really sticks out, its cause is immediately understood, which makes it a much better tool...
Thank you Charles Darwin, krivium, johnego, abraxalito and jacob2 for your contributions. I think our lack in the body of scientific knowledge and measurement in this particular field has given snake oil peddlers too much leeway. However, I will respond harshly against any response that hampers progress, no matter from whom it comes from, simply because it benefits snake oil peddlers. Hopefully the situation will become better in the future.
De facto.
Four out of four people (me aside) found the sound of SE 845 amp closer to the live musical event.
In other words, even it was a merely illusion, they have perceived a more realistic sound, not the live event but enough close.
Just for example they found very realistic and engaging the first movement of the Appassionata performed by Daniel Baremboim (DG redbook).
The song is very dynamic and contains very challenging transients.
All the four people are not audiophile, they don't know anything about audio, but they have a good musical education and they like attending to live events.
So the only goal for them was listening to the music they like as real as possible.
During the listening session only acoustic, classical and jazz music was played.
They don't care about psychoacoustic and measurements, so their approach is exclusively to the music they listen to.
Therefore they are absolutely not conditioned by our battles between measurements and listening tests, they do not even know its existence.
There was nothing to sell and nothing to buy, so no commercial conditioning.
If this is accepted as the fact I go beyond, otherwise I stop here.
What I mean is that there is no room for claims like "the 845 amp coloured the sound with its distortion" or "it was their subjective and personal taste" and so on.
In other words, if you think I should explain them that they were wrong with their evaluation because of the above reasons, audio reproduction does no longer make sense.
Therefore it would not make sense to use the measurements to prove the opposite of what was perceived by the listeners.
Audio systems are not a thing for a select few who attend forums, they are the mean used by ordinary people to listen to music.
And it must be pleasant rather than a torture.
In the end you can't impose your vision on other people, in the real world ordinary people buy an audio system to artificially reproduce the musical event at home in the way they feel most enjoyable, that is in the way they believe is closest to the real musical event.
The challenge
If all of the above is accepted, the challenge is very simple: find the right parameters with any measurement to understand what guided their choice.
Again, if the conclusion will be that people like distortion, that is.
The amplifier with more distortion is better than the one with less distortion, regardless of the usual and opposite interpretation of the measurements.
Now if someone expert on measurements is interested to undertake this test, please let me know what tools I need and what measurements I have to perform.
I will do my best to follow your indications.
Four out of four people (me aside) found the sound of SE 845 amp closer to the live musical event.
In other words, even it was a merely illusion, they have perceived a more realistic sound, not the live event but enough close.
Just for example they found very realistic and engaging the first movement of the Appassionata performed by Daniel Baremboim (DG redbook).
The song is very dynamic and contains very challenging transients.
All the four people are not audiophile, they don't know anything about audio, but they have a good musical education and they like attending to live events.
So the only goal for them was listening to the music they like as real as possible.
During the listening session only acoustic, classical and jazz music was played.
They don't care about psychoacoustic and measurements, so their approach is exclusively to the music they listen to.
Therefore they are absolutely not conditioned by our battles between measurements and listening tests, they do not even know its existence.
There was nothing to sell and nothing to buy, so no commercial conditioning.
If this is accepted as the fact I go beyond, otherwise I stop here.
What I mean is that there is no room for claims like "the 845 amp coloured the sound with its distortion" or "it was their subjective and personal taste" and so on.
In other words, if you think I should explain them that they were wrong with their evaluation because of the above reasons, audio reproduction does no longer make sense.
Therefore it would not make sense to use the measurements to prove the opposite of what was perceived by the listeners.
Audio systems are not a thing for a select few who attend forums, they are the mean used by ordinary people to listen to music.
And it must be pleasant rather than a torture.
In the end you can't impose your vision on other people, in the real world ordinary people buy an audio system to artificially reproduce the musical event at home in the way they feel most enjoyable, that is in the way they believe is closest to the real musical event.
The challenge
If all of the above is accepted, the challenge is very simple: find the right parameters with any measurement to understand what guided their choice.
Again, if the conclusion will be that people like distortion, that is.
The amplifier with more distortion is better than the one with less distortion, regardless of the usual and opposite interpretation of the measurements.
Now if someone expert on measurements is interested to undertake this test, please let me know what tools I need and what measurements I have to perform.
I will do my best to follow your indications.
Please search the forum, Earl has stated several times that with respect to the amplification chain it is a done deal. He finds numerous ordinary commercial amplifiers transparent. He has little interest in the sound of cables, magic stones, etc. at all. He would probably laugh at the thought of using a $$ SET amp with his speakers.
Last edited:
It was not a $$ SET, it was a chep chinese amp although heavily modified. 400 USD if I remember correctly, much cheaper than the Benchmark.
Last edited:
Most likely yes. But we can not be sure without measurement.
You have repeatedly stated non interest on the subject hence any explanation you'd be able to give would only be brushing the skin.
Last edited:
Measurement is one thing. Demonstrating convincing evidence of audibility can be the more difficult problem. We probably should be more concerned about the latter problem given the ultimate purpose of music reproduction for human consumption. Unfortunately, many engineers seems to believe that it only takes some engineering common sense to do it properly.
Last edited:
One problem is that the stereo illusion is based on the precise relationship between two channels. An assumption we can measure each channel individually and find out everything we need to know seems likely a conceptual error in terms of how we think about audio equipment measurements.
He's talking about speakers not amplifiers, the effects of the non-linearities are different.
Andrea_mori,
I see an issue with the test and the situation you describe: you don't have absolute reference for the test and the situation you describe is 'anecdotal evidence'.
About the test:
Ok your friends and you go to acoustic events but this test put to much emphasis on your auditive memory to be accurate or meaningfull.
We have poor long term auditive memory and short term is variable from one to another.
If you base your test only on memory you could at best define what your group prefer ( your taste, preference) but you could not have other meaningfull conclusion than this one.
Only way to validate your goal could be with real instrument played in front of you and then listen to the recorded event quickly after ( and if possible in a double abx configuration).
Even then you'll have to deal with the room and the mismatch in the way real instrument interact with it and the way your loudspeakers does it ( directivity mismatch between instruments and loudspeakers).
To be honest i've been part of this kind of test and this is a nightmare to set up and the results are not easy to read ( you have to make statistic analysis and to have a relevant number of participant, matched listening levels, etc, etc,...).
And as an outcome you could be very well desillusioned as it won't tell you anything about what parameters is of importance, only what you like/ are pleased with.
About anecdotal evidence be carefull about group effect. Even if you try not to be influenced you'll be...
I've got the exact inverse 'anecdotal evidence': one of my friend loaned his speakers to me.
He visited me and was blown away by the difference between the result in his home and mine: before even looking at the room and the setup he looked at amp and cd player and started to ramble about how much he spent in his gear...
He own a very nice Studer cd player, EAR tube preamp and a Kora Galaxy tube amp (2x50w). My chain was a cheap noname cd player directly through an Amcron/Crown D75.
When he brought back the loudspeakers i gave the d75 to test in his own room and setup and he was disgusted as outcome because sound was clearer, punchier more accurate... until i lowered the pre input on his EAR preamp ( two stage with pre and output volume) and the preamplifier triode input tube started to work in a more linear fashion...
He missed the 'body' the distortion of the EAR brought to the table though ( which i must reckon is very pleasing to the ear, but not accurate in any way in my view).
Only direct comparison ( with an absolute reference- his own amplifier in this case) brought him a smile as there was a difference between his 5k euros chain and mine (200euros) in his favour ( for his own preference, not mine. And i must say i don't see price as a meaningfull parameter in this kind of evaluation but it was his concern so...).
I see an issue with the test and the situation you describe: you don't have absolute reference for the test and the situation you describe is 'anecdotal evidence'.
About the test:
Ok your friends and you go to acoustic events but this test put to much emphasis on your auditive memory to be accurate or meaningfull.
We have poor long term auditive memory and short term is variable from one to another.
If you base your test only on memory you could at best define what your group prefer ( your taste, preference) but you could not have other meaningfull conclusion than this one.
Only way to validate your goal could be with real instrument played in front of you and then listen to the recorded event quickly after ( and if possible in a double abx configuration).
Even then you'll have to deal with the room and the mismatch in the way real instrument interact with it and the way your loudspeakers does it ( directivity mismatch between instruments and loudspeakers).
To be honest i've been part of this kind of test and this is a nightmare to set up and the results are not easy to read ( you have to make statistic analysis and to have a relevant number of participant, matched listening levels, etc, etc,...).
And as an outcome you could be very well desillusioned as it won't tell you anything about what parameters is of importance, only what you like/ are pleased with.
About anecdotal evidence be carefull about group effect. Even if you try not to be influenced you'll be...
I've got the exact inverse 'anecdotal evidence': one of my friend loaned his speakers to me.
He visited me and was blown away by the difference between the result in his home and mine: before even looking at the room and the setup he looked at amp and cd player and started to ramble about how much he spent in his gear...
He own a very nice Studer cd player, EAR tube preamp and a Kora Galaxy tube amp (2x50w). My chain was a cheap noname cd player directly through an Amcron/Crown D75.
When he brought back the loudspeakers i gave the d75 to test in his own room and setup and he was disgusted as outcome because sound was clearer, punchier more accurate... until i lowered the pre input on his EAR preamp ( two stage with pre and output volume) and the preamplifier triode input tube started to work in a more linear fashion...
He missed the 'body' the distortion of the EAR brought to the table though ( which i must reckon is very pleasing to the ear, but not accurate in any way in my view).
Only direct comparison ( with an absolute reference- his own amplifier in this case) brought him a smile as there was a difference between his 5k euros chain and mine (200euros) in his favour ( for his own preference, not mine. And i must say i don't see price as a meaningfull parameter in this kind of evaluation but it was his concern so...).
Last edited:
The difference would emphasize the fact that the number itself is meaningless, wouldn't it?
You are wrong, I am here to learn and I have given my total willingness to carry out all the measurements that an expert deems necessary.
This is absolutely clear from what I posted.
All right, I found measurement of amplifiers in isolation would not reflect acoustic output parameters in a linear fashion. So when comparing sound quality and measurement, the more relevant measurement would be those made on the acoustic output with all related room effect and reflection that comes with it. And most likely you made no such effort.
It was not a group test, one of the people had come to pick up the audio system that she still listen to today.
"He missed the 'body' the distortion ...."
Is this wrong?
It does not matter.
Would you impose your vision on other people?
Do you want to make the choice for them?
Does this mean that measuring a single component of the audio chain is useless?
- Status
- Not open for further replies.