Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: How do we establish "better/best"? (Read 44667 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

How do we establish "better/best"?

I am fairly sure that out of 100 people, most (if not all) would say a 44/16/stereo wav file of a contemporary/popular/easily-recognizable song does sound "better" than the same file at 11/8/mono.

But how is this established?

How is it that we know (or at least think we know) that in our car, music played on FM does sound signficantly better than music played on AM?

How is it that I can hear a song for the first time and make a subjective decision that 1) the playback equipment sounds bad or 2) the recording/mastering quality of the song is bad or 3) both?

How do we establish "better/best"?

Reply #1
giving a general short answer to a general question:

the fact that something for someone is/sounds/smells/(...) 'better' is because the person compares A to B and decides that because of factor a1, a2, etc A is better than B.
this lies in the nature of 'better', it's a comparison (better than, something cannot just be better per se. it is always different to something in regards to quality.

if somebody (P) would be aware of a certain input for the first time, let's say a (human) voice, then at first there is no better or worse. (the person might like it or may not because of similar non-voice stimuli that it compares it with but let's put that aside for now). then, when P gets to know some other voices from different people of different tones from the same person, etc, P will be able to compare them and decides what sounds better or worse.

let's further assume that compressed audio should reflect the reality as we see (or better hear) it most accurately to be 'good'. i.e. we compare compressed voices with non compressed voices and detect differences.
most of the time the compressed voice is labeled as worse, as it falls out of the pattern of voices we hear every day.

Quote
How is it that I can hear a song for the first time and make a subjective decision (about quality)
thesis:
- because you compare it to other auditive input of the category 'music' you know and decides: this one does not not reflect the (subjective) normal reality (i.e. other voices or music made and perceived  directly by other humans or by yourself).
different = potentially bad.

this is just a quick sum-up what I think about this and could/should be more elaborate and better structured than as it is now but I hope you get the idea.
Nothing but a Heartache - Since I found my Baby ;)

How do we establish "better/best"?

Reply #2
I think it can also be done with only science.  The human ear is a machine and can be understood given the time.  Ill make an analogy to our eyes.  There is a limit to how small of a pixel a person can see.  it has been measured.  this is for a given distance mind you.  but once an object is so small that you can not distinguish it from another it with different features (like different shapes) then it doesn't matter quite what shape it is does it? we havn't reached it yet but once pixels are small enough there will be no reason to go smaller. same with audio.

there might not be a resolution to the ear since its anologe but there is a resolution to our brain.  our eyes can see very small but our brain can't tell the difference.

whatever the specs are they can be measured and will be correct but people will debate them none the less.

How do we establish "better/best"?

Reply #3
Human beings can't really hear above ~20KHz or so (with RARE exceptions), so I'd say we've already reached the limits of audio. The dynamic range you can actually hear depends on how quiet the room is, though above a certain SPL, sound will damage your ears, so that sets an upper limit of what you should hear.

How do we establish "better/best"?

Reply #4
but how much resolution? this is hard because its really the brain... not our ears as they are not digital.  its like records.  its analog, it can capture things so minut that digital can never match that, but if our brain has a limit how minut it can tell the difference between then it could.

How do we establish "better/best"?

Reply #5
Resolution of what? Audio has a lot of aspects.
And digital can capture things -exactly- (not approximately, but exactly) with the right parameters, especially in regards to audio. That "analog can capture things so minute that digital can't" argument has been refuted a lot.

How do we establish "better/best"?

Reply #6
http://www.hydrogenaudio.org/forums/index....showtopic=40134
http://www.hydrogenaudio.org/forums/index....showtopic=47827
Here we go again, lol.

And just to stir the pot a little, practically speaking, quantization error does prevent capturing things exactly.

But to let people know where I stand (I won't pretend that this makes a difference), I believe that 44.1/16 is more than adequate for a final product.

How do we establish "better/best"?

Reply #7
an anolog system of anything has infinite resolution compared to any non quantum computing.  Take for example a needle on a pivot.  One end of the needle is pushing on some deformable material, clay for instance.  if you where to run that clay surface by the needle at a set rate and then wiggled the other end of the needle sporadically it would get a bunch of impressions.  now in digital how do you store those impression levels?  you would need an infinite number of values. 

this is why calculus was invented.  to be able to do math with infinite things.  but you can not use calculus do things machanical that are infinite, because in a given amount of time how are you going to run an infinite number of calculations?

How do we establish "better/best"?

Reply #8
but how much resolution? this is hard because its really the brain... not our ears as they are not digital.  its like records.  its analog, it can capture things so minut that digital can never match that, but if our brain has a limit how minut it can tell the difference between then it could.
Please stop trolling. If you really believe this then you need to do some background reading (maybe learn what the terms SNR and quantization noise mean, for a start) before you start a debate on it.
Quote
an anolog system of anything has infinite resolution compared to any non quantum computing. Take for example a needle on a pivot. One end of the needle is pushing on some deformable material, clay for instance. if you where to run that clay surface by the needle at a set rate and then wiggled the other end of the needle sporadically it would get a bunch of impressions. now in digital how do you store those impression levels? you would need an infinite number of values.
I'll give you the benefit of the doubt and assume that you are merely ignorant and not a troll.

Analog electronic systems do not have infinite resolution. Small changes in voltage are swamped by noise generated in all the conductors and semiconductors in the system. This noise is measured with a Signal to Noise Ratio (SNR) - when this is low enough the signal is effectively lost. Noise is the limiting factor in both analog and digital audio systems.

Quote
this is why calculus was invented. to be able to do math with infinite things. but you can not use calculus do things machanical that are infinite, because in a given amount of time how are you going to run an infinite number of calculations?
Eh? What exactly are you suggesting here?

How do we establish "better/best"?

Reply #9
The only way to determine better/best, IMO, is ABC/HR.

How do we establish "better/best"?

Reply #10

but how much resolution? this is hard because its really the brain... not our ears as they are not digital.  its like records.  its analog, it can capture things so minut that digital can never match that, but if our brain has a limit how minut it can tell the difference between then it could.
Please stop trolling. If you really believe this then you need to do some background reading (maybe learn what the terms SNR and quantization noise mean, for a start) before you start a debate on it.
Quote
an anolog system of anything has infinite resolution compared to any non quantum computing. Take for example a needle on a pivot. One end of the needle is pushing on some deformable material, clay for instance. if you where to run that clay surface by the needle at a set rate and then wiggled the other end of the needle sporadically it would get a bunch of impressions. now in digital how do you store those impression levels? you would need an infinite number of values.
I'll give you the benefit of the doubt and assume that you are merely ignorant and not a troll.

Analog electronic systems do not have infinite resolution. Small changes in voltage are swamped by noise generated in all the conductors and semiconductors in the system. This noise is measured with a Signal to Noise Ratio (SNR) - when this is low enough the signal is effectively lost. Noise is the limiting factor in both analog and digital audio systems.

Quote
this is why calculus was invented. to be able to do math with infinite things. but you can not use calculus do things machanical that are infinite, because in a given amount of time how are you going to run an infinite number of calculations?
Eh? What exactly are you suggesting here?




Did i say anything about an electrical system? i am merely talking about how to store something that is analog, just how to represent it.  I don't appreciate all the insults.

How do we establish "better/best"?

Reply #11
there might not be a resolution to the ear since its anologe but there is a resolution to our brain.  our eyes can see very small but our brain can't tell the difference.


There most certainly is a "resolution" limit to any analog, and there is certainly a well-established absolute threshold of hearing, one that is very close to, i.e. just above, the actual noise level due to molecules/atoms in the atmosphere bouncing off the eardrum (literally, yes, that's what "air pressure" is).

Furthermore, the ear reports its transduction to the brain in the form of pulse-position modulation, with quite limited time resolution, and not as good frequency resolution as you might expect.  30dB at a given level is about the best SNR you can expect from that part of the system, and that's well documented in the literature, too. Yes, Scharf says 40dB for tone masking noise in high barks, but that's because the barks are too wide at high frequencies, and you should use a decent ERB scale instead.

The best interaural time resolution reported anywhere is in the neighborhood of 5 microseconds, and more reports are in the 10-40 microsecond range (but how you test is very, very important and touchy here).

So, there are some fundamental limits known very well.

White noise at roughly 6dB SPL is the same level as the atmospheric noise level at your ear. Going much below that (for white noise) is pretty meaningless. -5.5 dB is the best reported "noise masking noise" report in the literature that I use.  More to the point, absolute threshold must take into account frequency and ERB/critical bandwidth, and it is just, barely, concievably possible that a completely unimpaired listener in the quietest possible room MIGHT JUST be able to hear that white noise in the 1kHz range near the ear canal resonance. Maybe.

So that 6dB SPL white noise limit looks pretty solid. We can take that as a decent lower level for anything ever required.

Let's go up 20 bits from that, we're at 126dB. That's louder than most any electronic reproduction system can manage, short or long term (I'm not talking about cars turned into pressure chambers for "loud stereo" testing here).

Even at 16 bits we get to 102dB. That's up there.

For 24 bits, we get to 150dB SPL. That's "instant damage to hearing" range.

Let's not even talk about 32 bit uniform PCM. That's military.

Did i say anything about an electrical system? i am merely talking about how to store something that is analog, just how to represent it.  I don't appreciate all the insults.



You need to slow down right now and realize that Quantum Mechanics, the size and charge of electrons, the mass of atoms, etc, puts a low level on any kind of analog system. These limits are not exceedling small, they are not something to be waved away, we can detect one photon with our eyes, and the noise level of the atmosphere I've addressed in my previous post.

Any form of analog capture of sound suffers from the issue of atmospheric noise. Sound does not exist, ever, in any form, WITHOUT said noise, it can not, as sound is defined as short-term variations in air pressure. If you have any air, you have noise.

Said level is easily detectable with instruments, and intrudes into all real systems, no matter what kind of analog they use for their storage, be it a time-continuous, frequency-continuous analog, or a discrete-time, discrete-level analog of the original signal.

The only way to determine better/best, IMO, is ABC/HR.


This is a separate question, and a very difficult one.

"Best" for one person may not be "best" for another. "Best" can mean "what the listener prefers", or "the most accurate" or presumably something else determined by the listener at the time of listening. If it's some kinds of especially "interesting" sound, "silence" may be the prefered sound.

But "best" is preference, and in the present world, only preference.

One can say "best capture of this point" or something like that, but until we capture an entire soundfield and reproduce it, there's nothing like an analytical "best" to be had.

"Best" is preference, really. The problem with the OP is that it is philosphically interesting more than it is technically interesting.
-----
J. D. (jj) Johnston

How do we establish "better/best"?

Reply #12
"Best" is preference, really. The problem with the OP is that it is philosphically interesting more than it is technically interesting.


In my experience, a vast majority (if not all) automobile listeners prefer music played on CD vs FM, and on FM vs AM.  This would appear to be technically measureable, as there are technical differences between CD playback and FM broadcasts & AM broadcasts.  So I disagree that it is relegated to impractical philosophy.  But I also think it is more than just statistical sciences, which discount human preference as arbitrary or illogical.

How do we establish "better/best"?

Reply #13
In my experience, a vast majority (if not all) automobile listeners prefer music played on CD vs FM, and on FM vs AM.  This would appear to be technically measureable, as there are technical differences between CD playback and FM broadcasts & AM broadcasts.  So I disagree that it is relegated to impractical philosophy.  But I also think it is more than just statistical sciences, which discount human preference as arbitrary or illogical.



You show that two measurements may correlate. No doubt.

Correlation, of course, does not imply causation. But, of course, it can.

Indeed, higher quality is undoubtedly a factor.  However, I think that singling out the "technical differences" has to be justified. I agree that some quality, convenience, and durability issues are certainly involved, but that it would require a lot more work to determine what is actually happening.

As to "impractical philosophy", please show me where I said "impractical philosophy". All philosophy is not impractical, in fact this very discussion shows its import in this very subject, I would suggest.  Finally, I do not agree that "statistical sciences" discount preference as arbitrary or illogical. Statistics makes  a measure of a preference, it neither validates nor discounts it.
-----
J. D. (jj) Johnston

How do we establish "better/best"?

Reply #14
The only way to determine better/best, IMO, is ABC/HR.
"Best" for one person may not be "best" for another. "Best" can mean "what the listener prefers", or "the most accurate" or presumably something else determined by the listener at the time of listening. If it's some kinds of especially "interesting" sound, "silence" may be the prefered sound.
Exactly. Which is why ABC/HR is the only way. Suppose you got 2 waveform files, using different encoding methods. Play them inside ABC/HR. Give a score to 1 track, another score to the other track. Find out which is the best according to your score.

My solution still stands

Note: I say ABC/HR not ABX there. Comprende?

How do we establish "better/best"?

Reply #15
The only way to determine better/best, IMO, is ABC/HR.
"Best" for one person may not be "best" for another. "Best" can mean "what the listener prefers", or "the most accurate" or presumably something else determined by the listener at the time of listening. If it's some kinds of especially "interesting" sound, "silence" may be the prefered sound.
Exactly. Which is why ABC/HR is the only way. Suppose you got 2 waveform files, using different encoding methods. Play them inside ABC/HR. Give a score to 1 track, another score to the other track. Find out which is the best according to your score.

My solution still stands

Note: I say ABC/HR not ABX there. Comprende?


Why do you need the hidden reference? You're asking "which of these two is better" not "which one sounds more like the original".

You can determine, for ONE person, what is "best" at a given time and place. I assert that's about all you can do in most tests.

If you test many people, you may get a variety of answers as to which is "best". Sometimes you'll get a concensus, very often you'll get several clusters of "best" that correspond to different kinds of preference.
-----
J. D. (jj) Johnston

How do we establish "better/best"?

Reply #16
Why do you need the hidden reference? You're asking "which of these two is better" not "which one sounds more like the original".
Let me quote the description from ff123's ABC/HR page:
Quote
Each group (there are eight groups) consists of two files: the reference file and a sample. The listener doesn't know which is which because the program has shuffled them within the group.

Now, if you want to know what's better, you can use just one group, assign (arbitrarily) one file as the reference, and the other file as the sample. You won't know! It's blind. Run the test, assign a score to both files, the one with the highest score wins.

Even guruboolez himself had, in a few (rare) occasion, mistook the sample for the reference, and assign a score to the reference instead... so it's an honest-to-goodness blind and scored test.

You can determine, for ONE person, what is "best" at a given time and place. I assert that's about all you can do in most tests.

If you test many people, you may get a variety of answers as to which is "best". Sometimes you'll get a concensus, very often you'll get several clusters of "best" that correspond to different kinds of preference.
I did not refute you on this. In fact, I agree with your position. "Best" for one is different from "best" for another.

I was only pointing out the tool by which one can determine which is "best" for him/her.

How do we establish "better/best"?

Reply #17
Quote
The only way to determine better/best, IMO, is ABC/HR.

Quote
Which is why ABC/HR is the only way. Suppose you got 2 waveform files, using different encoding methods. Play them inside ABC/HR. Give a score to 1 track, another score to the other track. Find out which is the best according to your score.

My solution still stands

Quote
I did not refute you on this. In fact, I agree with your position. "Best" for one is different from "best" for another.

I was only pointing out the tool by which one can determine which is "best" for him/her.
seems confuse and create a paradox.

you mean that someone really feel that "this" is best and using the "tool" for comparisons he find the "real" result.
if this "real" result is against his taste,what should he do?
he change his taste of what is best or trash the scientific result?

what the scientific result is prooving for his taste about what is "best/better"?



edit
typos

 

How do we establish "better/best"?

Reply #18
I am fairly sure that out of 100 people, most (if not all) would say a 44/16/stereo wav file of a contemporary/popular/easily-recognizable song does sound "better" than the same file at 11/8/mono.

But how is this established?

How is it that we know (or at least think we know) that in our car, music played on FM does sound signficantly better than music played on AM?

How is it that I can hear a song for the first time and make a subjective decision that 1) the playback equipment sounds bad or 2) the recording/mastering quality of the song is bad or 3) both?

44kHz recordings are more likely to be transparent (indistinguishable from live music).
16bit recordings are more likely to be transparent as well.
Stereo is nice because people have two ears.

You can compare it subjectively with live/uncompressed music; if you notice something unnatural in one method but not in another, then you can establish what's better.

How do we establish "better/best"?

Reply #19
seems confuse and create a paradox.

you mean that someone really feel that "this" is best and using the "tool" for comparisons he find the "real" result.
if this "real" result is against his taste,what should he do?
he change his taste of what is best or trash the scientific result?

what the scientific result is prooving for his taste about what is "best/better"?
What paradox?

I'm sorry to say this, but you seem to have not the slightest grasp of what ABC/HR is.

It is not an automated tool to tell you which one is better.

It is simply a comparator & scorer tool. It will present both audio files, you play them manually, and you score them manually, fully according to your taste. If you find the left one is better, give it a higher score. And vice versa.

It should be easy to deduce, based on how this tool works, that whichever scores better means that that one is better than the other, according to your liking

When the test concludes, ABC/HR will then 'uncloak' what the left and the right samples are. You just compare your score, and realize, "Oh, L has better score than R. That means, L is better for me than R."

Simple, no?

The beauty of ABC/HR is the blind scoring method it imposes; there is no such thing as codec bias etc.

For instance, you may be (unconsciously) preferring codec X instead of Z, because in your mind (or based on hearsay) Z introduces more white noise than X, X only having slight pre-echo. So you set out to prove it with ABC/HR.

As you run the ABC/HR session, you evidenly found out that L is more enjoyable to you than R. You can't really pin down the reason why, but you definitely like L better than R. So you conclude the test. Surprise, surprise, L is Z, R is X.

Based on ABC/HR you definitely prefer L than R. When the hidden samples are revealed, it becomes evident that you definitely prefer Z than X. To your ear, when tested blindly, Z sounds better than X.

ABC/HR is a great tool to dispel all codec bias.

How do we establish "better/best"?

Reply #20
Quote
It is not an automated tool ..
yes,i know.

Quote
Simple, no?
after your explanations.....yes! 

Quote
What paradox?
mine.
my questions was not only about ABC/HR but for the whole topic.

Quote
fully according to your taste
ok but taste is personal(individual)!
is easy to deduce but don't means that what is "better" is the "best" for my(our) taste.
as "best" is individual taste and "better" is the scientific result,we trash the scientific result or  change the mind as(if) "best/better" are differents?
worse: if we change the mind means that we don't have personal taste...(sorry,my poor english can't explain better than this)

ps:thanks for your complete and clever explantions about ABC/HR pepoluan! 

...but the "paradox" arise again(if you understood what i mean)

How do we establish "better/best"?

Reply #21
Ahhh, okay jlt, I got what you meant

Well, yes, there's a problem of my individual taste against the public conception/taste.

In that kind of conflict, there's no other way but to gather a statistics, i.e. we compare Z and X (as in my example above) across, say, the 30'000 member of HA. We gather the scores for Z and X, and do some statistical calculations (I'm not really good at statistics, so please forgive me) for standard deviation, median, etc. etc. etc.

... of course the best we can come up with is something 'diplomatic'  like:

Quote
According to our survey, there's 95% confidence in saying that more than 70% of HA members prefer Z over X (with a score of 4.5 against 3.2)

Of course the same survey also implies that almost 30% of HA members (with 95% confidence) prefer X over Z ...  ... but at least the clear majority (i.e. > 66%) prefers Z over X.

It's a bit like clothing situation... my favorite day-to-day wear is black polo shirt over black denims with black hiking shoes... and I have 95% confidence that less than 5% of the general population prefer that kind of color combination...

Ahh, statistics. There's lies, there's damn lies, and there's statistics. (Paraphrased from Benjamin D'Israeli)

How do we establish "better/best"?

Reply #22
Quote
and I have 95% confidence that less than 5% of the general population prefer that kind of color combination...
ah!ah!ah!... lol
you're too much.

you wrote the word explaining the meaning of "paradox": is conflict! (lol,my poor english really bore me because now came one new paradox(or conflict?):what is "better" and what is "best"?! conflict or paradox?!    )

thanks pepoluan,your last 'diplomatic' post show how the 'statistics' works.
I'm not really good at statistics too because the "ghost of paradox"(or conflict?) back "alive" sometimes. lol


How do we establish "better/best"?

Reply #23
For me better works by a simply comparison, if something A is better than something B in all aspects it can be considered better than A.
This is unlikely to happen, so A could be better than B in some aspects, so A is better then B at [ some aspects] but can't be considered better as a whole.
So far it's theoretically possible to be objective.

Here we have some aspects where B is better then A, some where A is better than B or both are equal in some of these. We need to weigh those aspects to conclude which one is better, and this is of course subjective.


To best: I hate it.
It often implies absolutism that is seldom valid.
To make a valid 'best' statement you usually need to limit it in several ways.
So one can state that something is best at a given time, out of the [comparable things] one knows, and most likely only in his opinion.
If someone simply states that something is the best, he usually automatically implies at least those three limitations, which could make it a valid statement, the problem is that no one else knows what he automatically implied or not implied.
Uhm, that was more the communication-problem... anyway, I think one can't validly state that something is the best without limiting in several ways.

How do we establish "better/best"?

Reply #24
I was only pointing out the tool by which one can determine which is "best" for him/her.


And, I'm pointing out that having a 'reference' in a pure preference test may be a mistake.

In general, in any test, you don't give the subject something will confuse them.
-----
J. D. (jj) Johnston