IPB

Welcome Guest ( Log In | Register )

2 Pages V  < 1 2  
Reply to this topicStart new topic
Need Clarification on TOS8
Wyld Stallyn
post Apr 4 2013, 15:01
Post #26





Group: Members
Posts: 13
Joined: 29-March 13
From: Fuldatal
Member No.: 107444



I would like to ask something regarding TOS8 too.

If I have understood correctly, the point of this TOS is to provide quantifiability and comprehensiblity in claims regarding the nature of audio reproduction.

I would be interested why waveforms and graphs are per se not allowed then. Is it because they are technically hard to accurately quantify without a significant economic investment (speculation on my part), making it more impracticable for the common forum member? Or is there some other underlying reason involved? Because I personally can not see what would speak against e.g. waveform comparisons as a means to show a perceptable or non-perceptable difference in audio reproduction, in particular when it adheres to certain standards to reduce visual exploitations such as weird cutting off, grotesquely magnified scales (like the graphs of the recent innerfidelity break-in tests visually showing differences of a noteworthy magnitude when in fact they border on the measurably relevant), etc. That is of course assuming all other things being equal.

This post has been edited by Wyld Stallyn: Apr 4 2013, 15:03
Go to the top of the page
+Quote Post
2Bdecided
post Apr 4 2013, 15:25
Post #27


ReplayGain developer


Group: Developer
Posts: 5103
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



Some audible differences are invisible, some inaudible differences are easy to see. This is true in some respect whatever graph or measurement you choose to use.

When a lossy codec (like mp3) is intentionally changing the signal, the best result is not the one that looks closest to the original, but one that sounds closest to the original. Because of the way human ears work, what the waveform (or spectrogram, or...) looks like does not match what the result sounds like very well at all. Hence, listening tests.

Even the best resourced professional does not use measurements to assess codecs. These listening tests aren't run for fun, or because they're cheaper than looking at the output on a scope, but because they're the only useful way of judging audio quality in the context of codecs which intentionally change the audio signal without (hopefully) audibly degrading it...
http://mp3-tech.org/programmer/docs/w2006.zip

Cheers,
David.

This post has been edited by 2Bdecided: Apr 4 2013, 15:26
Go to the top of the page
+Quote Post
Wyld Stallyn
post Apr 4 2013, 15:46
Post #28





Group: Members
Posts: 13
Joined: 29-March 13
From: Fuldatal
Member No.: 107444



My OpenOffice copy isn't able to open the provided document.

I don't really understand what you are saying. Are you talking about influences of HRTF and uh, I forgot the term for this, uhm, the influence of the room? tongue.gif

That seems to me to be completely irrelevant for showing a difference between two signals though. If both signals have been recorded under the exactly same circumstances, if there is any noteworthy difference in the waveform then that should mean that there is a notable difference caused by the different component or settings, no matter how much that signal gets distorted by physics. If physical effects would cause a difference of +5 dB at a certain frequency in one circumstance and -8 in the other, would not the very same differences be applied to the altered signal on top of the alteration, making the alteration still equally perceptable (if it is in the realm of perceptability at all of course, we could argue that no claims for this can earnestly be made below a certain point that is not relevant to the average listener, who is the one going to be influenced into purchasing the product after all) due to the preserved difference in both circumstances? This should be the case, otherwise even ABX testing in a specific circumstance could not provide the slightest hint of orientation for different circumstances either, and it'd probably mean that there are no general physical patterns to rely upon, which is evidently not the case.

Or is it the case that other factors are influenced that are not representable with a waveform? Wouldn't just providing the relevant measurements satisfy our needs then? If need be maybe all sorts of measurements that can possibly be made, even if that seems like a logistical nightmare to me where I'd rather conduct an ABX test. Of course, what I would prefer to do for my nerves' sake is not the matter here.

This post has been edited by Wyld Stallyn: Apr 4 2013, 15:48
Go to the top of the page
+Quote Post
Porcus
post Apr 4 2013, 16:02
Post #29





Group: Members
Posts: 1842
Joined: 30-November 06
Member No.: 38207



I really don't understand what you are saying either, so I might have misinterpreted:

QUOTE (Wyld Stallyn @ Apr 4 2013, 16:46) *
If both signals have been recorded under the exactly same circumstances, if there is any noteworthy difference in the waveform then that should mean that there is a notable difference caused by the different component or settings, no matter how much that signal gets distorted by physics. If physical effects would cause a difference of +5 dB at a certain frequency in one circumstance and -8 in the other, would not the very same differences be applied to the altered signal on top of the alteration, making the alteration still equally perceptable


If what you are suggesting is the following:

- Suppose that two operations A and B transforming signal #0 into signals #a and #1b are such that #a and #b are not audibly distinguishable
- If the same transformation C is applied to both the latter, resulting in #ac and #bc, then since #a and #b were transparent, so will #ac and #bc also be

... then you are certainly wrong. (Counterargument: make an EQ or a volume change; scale it up so that the alteration and the negative one are both inaudible, but one is very close to; then #a is #0 plus a delta, and #b is #0 minus the same. Now let C be the reverse operation of one of them. Operation C will amplify one transformation and annihilate the other.)


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post
Wyld Stallyn
post Apr 4 2013, 16:30
Post #30





Group: Members
Posts: 13
Joined: 29-March 13
From: Fuldatal
Member No.: 107444



This may seem a bit dumb, but I actually didn't understand this part, you may want to blame it on the fact that I am not native, and I sure hope that's all there is to it:
"make an EQ or a volume change; scale it up so that the alteration and the negative one are both inaudible, but one is very close to;"
What do you mean with scaling up exactly? tongue.gif What do you mean with the negative? As in inverted signal? And very close to what?

Luckily understanding that part was not vital to your argument and I see what you mean, at least I think so.

However, my argument is this:
Have Signal a. Transform it so that you attain a1 which is audibly distinguishable. Now in two different settings A and B which introduce individual transformations that may or may not be noticable, let both a and a1 be measured. Signal a1 will be distinguishable from a in both settings to the same degree.

I guess my notation is formally incorrect, but hey.

This post has been edited by Wyld Stallyn: Apr 4 2013, 16:39
Go to the top of the page
+Quote Post
greynol
post Apr 4 2013, 16:32
Post #31





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



We don't hear with our eyes.

/enddiscussion


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
2Bdecided
post Apr 4 2013, 16:43
Post #32


ReplayGain developer


Group: Developer
Posts: 5103
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



First read and understand this...
http://en.wikipedia.org/wiki/Auditory_masking

...then take a quick look as this...
http://en.wikipedia.org/wiki/Psychoacoustics#Software

...then try to find a graph or measurement that indicates which elements of a complex audio signal are audible to the human ear, and which bits are not.

Hint: there is no such graph or measurement. The closest you get is with simulations of human hearing. PEAQ is an internationally standardised model of human hearing, that is trained and based upon the results of... listening tests!

Cheers,
David.
Go to the top of the page
+Quote Post
Wyld Stallyn
post Apr 4 2013, 16:46
Post #33





Group: Members
Posts: 13
Joined: 29-March 13
From: Fuldatal
Member No.: 107444



QUOTE (greynol @ Apr 4 2013, 17:32) *
We don't hear with our eyes.

/enddiscussion

I'm sorry but that is not the end of the discussion. Waveforms serve as a translation of the audible part of the frequency spectrum into the visible part, but that does not negate the difference in audible frequency if two methodically correctly measured signals differ from each other on waveforms. Only where a difference is measured will a difference be plotted, unless the equipment is fatally flawed.
QUOTE (2Bdecided @ Apr 4 2013, 17:43) *
First read and understand this...
http://en.wikipedia.org/wiki/Auditory_masking

...then take a quick look as this...
http://en.wikipedia.org/wiki/Psychoacoustics#Software

...then try to find a graph or measurement that indicates which elements of a complex audio signal are audible to the human ear, and which bits are not.

Hint: there is no such graph or measurement. The closest you get is with simulations of human hearing. PEAQ is an internationally standardised model of human hearing, that is trained and based upon the results of... listening tests!

Cheers,
David.

Ah yes, now I understand. A measurable difference may not be audible even if in theoretically audible dimensions because the affected frequencies may still be masked by others more prominent ones. Auditory masking was a concept I knew about (even if I understand it only roughly), but I failed to think of it. Thanks for bringing it more to my attention!

This post has been edited by Wyld Stallyn: Apr 4 2013, 16:52
Go to the top of the page
+Quote Post
greynol
post Apr 4 2013, 18:01
Post #34





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



QUOTE (Wyld Stallyn @ Apr 4 2013, 08:46) *
Waveforms serve as a translation of the audible part of the frequency spectrum into the visible part

You were already told(!), differences that may be made demonstrated visually do not necessarily translate into differences that can be detected audibly.

That you may question it doesn't change the reality of the situation, and will certainly not change the rule or the way it is enforced.


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
pdq
post Apr 4 2013, 18:03
Post #35





Group: Members
Posts: 3392
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



My interpretation of TOS #8 is this:

If a claim is made that something is audibly different then that claim must be backed up in a way that discerns audible differences, i.e. DBT listening tests.

OTOH, if no claim of an audible difference is made then you can use whatever measurements and graphs you like.

If you have properly supported a claim of audible difference, then measurements and graphs may be used to help explain why there is an audible difference.
Go to the top of the page
+Quote Post
Wyld Stallyn
post Apr 4 2013, 18:54
Post #36





Group: Members
Posts: 13
Joined: 29-March 13
From: Fuldatal
Member No.: 107444



QUOTE (greynol @ Apr 4 2013, 19:01) *
QUOTE (Wyld Stallyn @ Apr 4 2013, 08:46) *
Waveforms serve as a translation of the audible part of the frequency spectrum into the visible part

You were already told(!), differences that may be made demonstrated visually do not necessarily translate into differences that can be detected audibly.

That you may question it doesn't change the reality of the situation, and will certainly not change the rule or the way it is enforced.

Yes. I refer you to my last response. Up to that point it has in fact not been sufficiently well laid out.
Go to the top of the page
+Quote Post
greynol
post Apr 4 2013, 19:03
Post #37





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



It has been sufficiently well laid out in countless discussions over the years. I see no reason for anyone to have to go through it again when the forum has a search function.


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post
dhromed
post Apr 5 2013, 09:14
Post #38





Group: Members
Posts: 1303
Joined: 16-February 08
From: NL
Member No.: 51347



QUOTE (2Bdecided @ Apr 4 2013, 16:25) *
Some audible differences are invisible, some inaudible differences are easy to see. This is true in some respect whatever graph or measurement you choose to use.

Go to the top of the page
+Quote Post
Mach-X
post Apr 6 2013, 01:25
Post #39





Group: Members
Posts: 269
Joined: 29-July 12
From: Windsor, On, Ca
Member No.: 101859



I myself have wondered how this applies to statements about loudspeakers or headphones. I think its a given assumption unlike amplifiers or dacs that they do sound different. But when I attempted once to describe the difference between some loudspeakers I was auditioning, some members indicated I was violating tos 8. I did ask what was an appropriate way to quantify speaker qualities but didnt really get an answer.
Go to the top of the page
+Quote Post
Mach-X
post Apr 6 2013, 01:32
Post #40





Group: Members
Posts: 269
Joined: 29-July 12
From: Windsor, On, Ca
Member No.: 101859



Otoh, tos8 is fairly simple. Even established facts such as lossless is lossless is open to challenge, IF you can provide evidence and samples to prove it. Even if the claim is proven false it can help users to diagnose possible issues with their equipment or developers to iron bugs in encoders/decoders. Substantiate your claim with blind test data. Its that simple.
Go to the top of the page
+Quote Post

2 Pages V  < 1 2
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 27th August 2014 - 20:45