Can You Trust Your Ears? By Tom Nousaine

[Thomas savage]

I think you had some kind of error in your digital rendition of the analog visual color spectrum there -- those blacks don't appear to be as black as they should be.

Did you use a lossy image format?

Well I’m on a iPad so what ever the mishap at least it’s not amirs fault ...

I’m betting amirs going to chose the brown one, can’t speak for @Arnold Krueger dont know him well enough but I’d suggest the green one so it contrasts nice and we can clearly determine who’s getting the better of who.

 

[Thomas savage]

I don't think that accusation sticks at all. Thomas carries such purses everyday so is intimately familiar with how they look. His instagram page is full of such things.

Those colourful offerings I was carrying about round your gaff a few weeks back were in-fact my spare underpants .. it’s all the rage in Paris this season.

Anyhow, carry on

 

[Kal Rubinson]

Sounds interesting, but never even heard of it. Where / how is it implemented?

Implemented only in Meridian products. They have a white paper (although it may be brown at the edges by now).

 

[Kal Rubinson]

If I wanted to listen to stereo recordings (since they're the default content) with 3 channels, would the center channel simply be the same of the left and right?
Or do I need something that is natively 3 channel to get the desired effect?

I believe the latter is the only suitable way to do it. There are many, many other ways.

 

[svart-hvitt]

Putting it all together in what way?

Full disclosure: I respect Genelec, but I've never liked their monitors. I'm an ATC / Dynaudio guy.

Putting it all together: DAC, ADC, amplifiers, DSP incl. room correction for up to 30 speakers/bass system, cloud based calibration and customer support services. The speakers measure flat; neutrality is a safe choice. Hard to tell my 8351a 3-way coaxial apart from for example the Kii Three when you switch between them in the same room.

 

[fas42]

Define "high enough". Until you do, you're simply engaging in tautology.

Which is why I'm happy to keep putting the term in quotes ... an aural situation can be created, which at least a siginificant people have experienced, as shown by their understanding of descriptions of the subjective impression of it - they "know what I'm talking about". Thus, it does exist, for some subset of listeners - each person has their own explanations, theories about what is happening, and what is necessary to achieve it.

It's something like the "sound barrier" for planes - at that time, could anyone define "high enough" capability of the aircraft that would ensures a new design would pass it with ease, when people were still struggling to make it happen ... it's a time of learning, not one of formalising a well accepted situation.

 

[fas42]

Sidenote: easy method for getting "immersive" sound is to throw more amplifiers into the mix. This can be done in numerous ways: MCH, active speakers, centre channel, etc, etc. The common theme is, yes, more amplifiers, driving speakers wherever you wish to place them ... why this works is because each amplifier has less of a load, getting the job done of driving its associated speaker(s) to the required volume.

It's as simple as that ...

 

[fas42]

@fas42 we've been down this rat hole before. Until you can quantify what is "high enough standard" using data, this is nothing better than "I know it when I hear it", which makes it out of the bounds of science.

When something is experimental, that's how it works - how strong should a radio transmitter be in some area? Well, strong enough that that all relevant people can pick up the signal! OK, describe precisely the situation that every person could be in, the precise topography of the area, every atmospheric situation that could arise, every type of receiving device that may be used ... science doesn't have much useful to say here - and engineering just says, just make it plenty strong! Then we will check out what the feedback is, and move on from there ...

 

[fas42]

I play trombone and bass (acoustic and electric) and find myself doing that with genres that I actively played in (jazz, classical, blues, a bit of funk). Sometimes I seem to even mentally fill in progressions before they happen, then get shocked when the artist goes in a different direction. I also think I mentally fill in sounds I can't necessarily actually hear.

Interestingly, for genres for in which I have little to no hands-on experience (classical Chinese or Japanese, gamelan, EDM, liturgical chant) I don't find myself "listening as a musician".

I am completely hopeless, at any instrument, so I have to rely totally on what comes out of the speakers to tell me the story. It's the "not listening as a musician" that gives the game away, as to whether a system is "good enough" - a piece that you have no experience of, in a genre that you 'dislike' or have zero interest in - can you grok what is going on as a musical event; or is your whole being "fighting" the sound, being disturbed by it?

 

[Sal1950]

gamelan, EDM, liturgical chant

Huh, what, are you speaking English?

 

[watchnerd]

Huh, what, are you speaking English?

 

[j_j]

We see again MOS style scoring of preference (from 0 to -4). No ABX, forced binary choice here.

Two points,

1) ABC/hr, which is what those tests were, does involved a forced choice, in that one is obliged to identify one of B or C as the same as 'A'.

2) It's not an MOS (mean opinion score) scale, it is rather a difference scale. This is not merely semantics, difference scales have been shown clearly to work better in terms of subject consistency than preference (MOS is a preference scale) scales.

We used ABX (also pairwise AB before we had the hardware) to determine various threshold calculation behaviors before the ABC/hr setup was established.

 

[j_j]

Which is why I'm happy to keep putting the term in quotes ... an aural situation can be created, which at least a siginificant people have experienced, as shown by their understanding of descriptions of the subjective impression of it - they "know what I'm talking about".

You are now describing an issue of personal preference. You have yet to show, for instance, than an acceptable, reliable perception of front center distance, beyond (right in front) can be established. You have yet to show that palpable envelopment, including head movement, can be demonstrated.

In short, you are now saying you've hard things people like. So have I. That doesn't change the basic science of the issue. Sorry.

You need a falsifiable, repeatable claim, with metrics for distance, image location, envelopment, depth, you name it.

 

[j_j]

why this works is because each amplifier has less of a load, getting the job done of driving its associated speaker(s) to the required volume.

It's as simple as that ...

Since many blameless amplifiers exist these days, including ones that operate at truly insane (in terms of loudspeaker power handling and human hearing safety) power levels, I don't think it is even remotely "as simple as that". You have, at least, formed a working hypothesis, but not yet one that is falsifiable. How will you make your assertion falsifiable?

Yes, science is hard. I do it for a living.

 

[j_j]

I believe the latter is the only suitable way to do it. There are many, many other ways.

Oh (*&(*& yes, material made for 3 channel, PROPERLY made for 3 channel, is the only suitable way, and yes, there are many other options indeed. (shudder)

 

[amirm]

Two points,

1) ABC/hr, which is what those tests were, does involved a forced choice, in that one is obliged to identify one of B or C as the same as 'A'.

2) It's not an MOS (mean opinion score) scale, it is rather a difference scale. This is not merely semantics, difference scales have been shown clearly to work better in terms of subject consistency than preference (MOS is a preference scale) scales.

We used ABX (also pairwise AB before we had the hardware) to determine various threshold calculation behaviors before the ABC/hr setup was established.

I don't understand the nature of this back and forth JJ. ABX in no way or form provides any degree of impairment. Almost every codec test many of which I have provided sample of, provide such scores. As we both know well, that information is completely absent in ABX tests, yet fundamental to testing of lossy codecs. Track to track, the fidelity differences vary as does with bit rates. As does listener feedback. How can ABX be at all a useful tool here?

Can you give me any published lossy codec test that relied on ABX tests?

I am sure you also know that countless tests are done sighted in the process of development of codecs. It is not at all the dreamworld that Arny assumes.

 

[amirm]

Here is the Wiki on ABC/hr from HA forum: http://wiki.hydrogenaud.io/index.php?title=ABC/HR

"ABX is used to detect audible differences blindly, thereby removing personal bias or the placebo effect and, over multiple trials, estimating the probability that the tester was guessing.

While an ABC/HR tool can do the same, it adds to that the ability to provide a quality rating on a standardized scale, and multiple participants' results can be statistically evaluated to estimate error bars and statistical significance of differences between encoders or encoder parameters in ranking their quality or tying them.

ABC/HR tends to find particular application in low-to-medium bitrate listening tests (below the quality expected to constitute transparency). As part of a well-designed listening test, useful quality comparison can be made between a selection of encoders plus a high-quality anchor (high anchor) and low-quality anchor (low anchor) without the tester being aware of which encoder is being evaluated at any time.

Standardized quality or impairment scale[edit]
It is most common to use the 1.0 to 5.0 scale defined by ITU-R BS.1116. Any value (including fractions) between 1.0 and 5.0 is valid, with the exact whole number representing the following definitions to describe that degree of impairment:

• 5.0 : Imperceptible
• 4.0 : Perceptible, but not annoying
• 3.0 : Slightly annoying
• 2.0 : Annoying
• 1.0 : Very annoying"

When one is giving that grade there is nothing forced about it as there is in ABX. The role of the hidden reference and the way it gets graded is an entirely different animal than ABX testing.

Just because two things have wheels, it doesn't make an airplane the same as a car.

 

[j_j]

I don't understand the nature of this back and forth JJ. ABX in no way or form provides any degree of impairment. Almost every codec test many of which I have provided sample of, provide such scores. As we both know well, that information is completely absent in ABX tests, yet fundamental to testing of lossy codecs. Track to track, the fidelity differences vary as does with bit rates. As does listener feedback. How can ABX be at all a useful tool here?

Can you give me any published lossy codec test that relied on ABX tests?

I am sure you also know that countless tests are done sighted in the process of development of codecs. It is not at all the dreamworld that Arny assumes.

My comment isn't about ABX testing, per se, it's about ABC/hr testing, which is what the codec tests you cited were. The distinction between "MOS" and "difference" is extremely important. They are not "MOS" tests. MOS tests lead rapidly to all sorts of confounding issues. Even impairment tests do that. Pure distance tests work the best with human subjects.

It is also important to realize that ABC/hr requires a forced-choice decision. It does have the ability to determine, given sufficient data, if a subject is performing beyond probability, merely by the subjects ratings for B and C, one of which is obliged to be identified as the same as 'A'.

Neither of these makes it an ABX test.

To my knowledge, there is not yet a known, published codec that can pass an ABX test, although that would be a worthy goal for codec designers.

 

[j_j]

Standardized quality or impairment scale[edit]
It is most common to use the 1.0 to 5.0 scale defined by ITU-R BS.1116. Any value (including fractions) between 1.0 and 5.0 is valid, with the exact whole number representing the following definitions to describe that degree of impairment:

• 5.0 : Imperceptible
• 4.0 : Perceptible, but not annoying
• 3.0 : Slightly annoying
• 2.0 : Annoying
• 1.0 : Very annoying"

When one is giving that grade there is nothing forced about it as there is in ABX. The role of the hidden reference and the way it gets graded is an entirely different animal than ABX testing.

Just because two things have wheels, it doesn't make an airplane the same as a car.

You are obliged by the test structure to give one of B or C a rating of 5. That is part of the test protocol. The test hardware is most often arranged to require it.

That, in and of itself, is a forced choice. The results of this forced choice can (and are) used in the same way as those resulting from an ABX test. The choice of "reference" is forced.

 

[amirm]

That, in and of itself, is a forced choice. The results of this forced choice can (and are) used in the same way as those resulting from an ABX test. The choice of "reference" is forced.

We are back to calling everything with a wheel a car. Here is an ABX output of a test I have run (Arny's no less):

foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/06/20 08:40:02

File A: C:\Users\Amir\Music\Arny's 30 Hz Jitter File\Arny's new files\no jitter.wav
File B: C:\Users\Amir\Music\Arny's 30 Hz Jitter File\Arny's new files\30 Hz noticable jitter 0.0125.flac

08:40:02 : Test started.
08:41:22 : 01/01 50.0%
08:41:46 : 02/02 25.0%
08:42:07 : 03/03 12.5%
08:42:26 : 03/04 31.3%
08:42:49 : 03/05 50.0%
08:43:15 : 04/06 34.4%
08:43:46 : 05/07 22.7%
08:44:26 : 06/08 14.5%
08:44:55 : 07/09 9.0%
08:45:34 : 08/10 5.5%
08:45:58 : 09/11 3.3%
08:46:28 : 10/12 1.9%
08:47:09 : 11/13 1.1%
08:47:35 : 11/14 2.9%
08:48:02 : 12/15 1.8%
08:48:28 : 13/16 1.1%
08:48:56 : 14/17 0.6%
08:49:49 : 14/18 1.5%
08:50:20 : 15/19 1.0%
08:50:46 : 16/20 0.6%
08:50:49 : Test finished.

----------
Total: 16/20 (0.6%)

---

Here is yet another example of fidelity based codec testing: http://www.aes.org/e-lib/browse.cfm?elib=16292

There is no similarity there. When I am taking an ABX test I am forced to make an A or B selection. No such thing goes on in these other tests where I am required to give a score.

Just because both tests are blind we don't call them the same. Likewise the fact that I have to provide a quality rating doesn't make it the same as binary choice that ABX has. Different animals altogether.