Home Theater Forum and Systems banner

1 - 20 of 27 Posts

·
Banned
Joined
·
1,131 Posts
Discussion Starter #1
AV Sound Lab said:
As with all disciplines in science, audio has a dedicated publication, the Journal of the Audio Engineering Society (JAES). While the JAES deals with other forms of research besides perceptual research, this article will focus primarily on perceptual research. The JAES and the contributing researchers have greatly advanced the understanding of how humans hear, what is preferred and, by association, loudspeaker design.

In its earliest form, perceptual research was used to determine what volumes were perceived to be the same across the audible frequency spectrum. [1]. As time progressed and equipment became more reliable, more advanced testing was possible. Such testing was able to determine the audibility of frequency response anomalies [2] and further research established audibility thresholds of distortion in various devices [3]. Later research was even able to correlate loudspeaker measurement to listener preference [4, 5]. This research has advanced so much that mathematical models have been created which accurately correlate (r=0.995) listener preference to measured loudspeaker performance [6]...

Read more.
This thread is for all discussion of the Perceptual Research article found on AV Sound Lab.
 

·
Registered
Joined
·
1,249 Posts
I still read over that last article you sent me regarding perceptual research, and having read the above article, am lead to one major question. If a speakers quality can be appraised by the human ear just as effectively (as the above article states) as measurements with expensive gear, then why do we bother? why not just fiddle around with our systems until they sound the best and then leave them at that? I sumise that we don't trust ourselves and hence rely on the measurements to reassure us that the money and time we have commited to our systems is justified and not just wishfull thinking or, as the above article puts it, bias toward a particular brand or style. Eitherway if we keep treading this path we will have psychological reasons for justifying psychoaccoustical annomilies that require a psychological bias to hear.

I think if it sounds good it sounds good, if it doesn't it doesn't, I fail to see how measuring a speaker will change ones accoustic preference.

Cheers
Dr F
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #3
The main problem is one of bias. The human ear is a superb measuring device in blinded circumstances, but it has been shown time and time again that when the listening tests are done in sighted situations factors such as quality correlate far less with preference than brand name and perceived status do.

As far as measuring speakers changing preference, you are right there is no relation. Rather, the point of this is that loudspeaker measurement has a direct relation to preference of individuals with normal hearing. So with proper knowledge of what measurements are subjectively superior one can build a speaker that sound even better than good.
 

·
Registered
Joined
·
1,249 Posts
So with emperical scientific data to support all this, a company can release a product with evidence of its accoustic quality that no-one can deny regardles of the brand or price tag. This will unsettle a lot of audiophiles and rattle a few snake oil cages. It hopefully will also make some people rethink their views on cheap brands like behringer and alto in the pro audio.

I have noticed of late that there has been a certain amount of backpedaling on DIYAUDIO.com about certain "neccesities" one needs in amplifiers and the sudden relunctance to use words like "musicality" when defining audio gear. I wonder if it is due in part to the lastest perceptual research?
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #5
So with emperical scientific data to support all this, a company can release a product with evidence of its accoustic quality that no-one can deny regardles of the brand or price tag. This will unsettle a lot of audiophiles and rattle a few snake oil cages. It hopefully will also make some people rethink their views on cheap brands like behringer and alto in the pro audio.
That would certainly be nice, but mass acceptance and more importantly understanding of this research must be had before this will ever happen.

The other factor is the room, a speaker can be technically perfect and if put in the wrong room or even improperly within a room it will sound terrible. This is, of course, covered in the relavent research as a majority of actual sound heard is not produced by the speaker, but rather is reflected in the room*.

*This presumes a reasonable sized home room and not a massive theater.

I have noticed of late that there has been a certain amount of backpedaling on DIYAUDIO.com about certain "neccesities" one needs in amplifiers and the sudden relunctance to use words like "musicality" when defining audio gear. I wonder if it is due in part to the lastest perceptual research?
I don't follow that forum at all so I cannot speak to this subject. I do know that there is a large amount of literature on audibility differences between components. Some of this research might be new, but the scientifically proven idea that well designed components used within their limits have no effect on sound quality (unless intentional) is not a new one. Perhaps it is just an adoption of this thought philosophy.
 

·
Registered
Joined
·
373 Posts
...evidence of its accoustic quality that no-one can deny regardles...
Humanity has infinite capacity to deny and marketing departments have large budgets to bias you in one direction or another.

The good news is that once you understand what the measurements tell you, at least you can make better decisions on what to buy for yourself.
 

·
Registered
Joined
·
2,431 Posts
I mark all of my CD's with a green marker as it helps prevent jitter and tightens the lower mids. This along with elevating all of my cables off of the floor has produced a significant impact on the tonal clarity of my system. :rofl2::rofl:
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #8
The good news is that once you understand what the measurements tell you, at least you can make better decisions on what to buy for yourself.
This is very true, but unfortunately properly correlating measurement to listener preference is not an easy skill to attain. Due to the complex interaction of variables this is something that needs to be practiced in bias minimized circumstances alongside deep review of relevant literature.

I mark all of my CD's with a green marker as it helps prevent jitter and tightens the lower mids. This along with elevating all of my cables off of the floor has produced a significant impact on the tonal clarity of my system. :rofl2::rofl:
These are perfect examples of bias completely confounding actual reality and experience. Such bias is seen virtually everywhere: audio, medicine, food, clothing etc...
 

·
Registered
Joined
·
393 Posts
Humanity has infinite capacity to deny and marketing departments have large budgets to bias you in one direction or another.

The good news is that once you understand what the measurements tell you, at least you can make better decisions on what to buy for yourself.
The other problem is: Are you sure the numbers on the spec sheet are the numbers that actually came out of the lab, or did marketing get hold of them and "tweak" them first? Were they measured to industry standards? Was all the equipment in calibration and correctly used? When they say 300Wpc, do they mean RMS, Peak, PMPO, or what? And was that measured from 20-20kHZ, or just at 1kHz? There's no government or industry body to check all the specs, so you kind of have to take them with a grain of salt.
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #10
The other problem is: Are you sure the numbers on the spec sheet are the numbers that actually came out of the lab, or did marketing get hold of them and "tweak" them first? Were they measured to industry standards? Was all the equipment in calibration and correctly used? When they say 300Wpc, do they mean RMS, Peak, PMPO, or what? And was that measured from 20-20kHZ, or just at 1kHz? There's no government or industry body to check all the specs, so you kind of have to take them with a grain of salt.
This is exactly why credible third party measurements such as Ilkka's are a requirement for proper analysis of any item when correlating measurement to perception.
 

·
Registered
Joined
·
373 Posts
This is very true, but unfortunately properly correlating measurement to listener preference is not an easy skill to attain. Due to the complex interaction of variables this is something that needs to be practiced in bias minimized circumstances alongside deep review of relevant literature.
I'm sure its no more simple or difficult t learn than many other things. In my mind, the biggest barrier to learning and application is access to relevant data.

This in itself makes a case for the need to audition the speakers you choose. As I remember it, Dr. Toole felt it was a rather simple short process to train the listener on what to listen for and how to describe what one is hearing.
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #12
This in itself makes a case for the need to audition the speakers you choose. As I remember it, Dr. Toole felt it was a rather simple short process to train the listener on what to listen for and how to describe what one is hearing.
The problem with simply auditioning speakers is that bias is so prevalent in the human experience. Toole and Olive conducted a research study where a panel of listeners rated speakers in sighted circumstance and then repeated the experiement with a blinded situation. The results were not surprising, the sighted situation was plagued with confounds: e.g., loudspeakers with high reputations or prices being rated at the top. While such effects were expected the authors were surprised at the magnitude of this effect. Once the bias was accounted for the preference ratings had no correlation with perceived status of the loudspeaker, but rather actual performance. The end result being: "The bottom line: if you want to know how a loudspeaker truly sounds you would be well advised do the listening tests 'blind'" [1]. This is just a single reason why simple listening tests performed by most are not sufficient to fully judge loudspeaker performance.

As far as the belief that training is or is not a long process that was dealt with directly in a paper by Olive [2]. The end result was that trained listeners more quickly arrive at a conclusion about the loudspeakers being auditioned while being more critical and having less variability. In contrast, untrained listeners come to the same conclusions more slowly and with higher variability. Additionally, untrained listeners became more reliable [or became self trained] over multiple trials.

References:

[1] Toole, Floyd E.; Olive, Sean E. Hearing is Believing vs. Believing is Hearing: Blind vs. Sighted Listening Tests, and Other Interesting Things. J. Audio Engineering Soc., Conference Paper 3894. November 1994/

[2] Olive, Sean E. Differences in Performance and Preference of Trained vs Untrained Listeners in Speaker Tests. J. Audio Engineering Soc., Vol. 51, No. 9. Pages 806 - 825. September 2003.
 

·
Registered
Joined
·
393 Posts
This is exactly why credible third party measurements such as Ilkka's are a requirement for proper analysis of any item when correlating measurement to perception.
Right, I kinda left that out of what I said, but it was (to me) implied.

Knowing what all the measurements, graphs, etc., mean is great, but it doesn't necessarily follow that you are getting good information from them.

You could probably include yourself in the "credible third party measurements". Or do you blush easily?
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #14
Knowing what all the measurements, graphs, etc., mean is great, but it doesn't necessarily follow that you are getting good information from them.
That is exactly why it is important to not only understand what the measurements mean, but how they are taken and/or derived so that proper judgments of credibility can be made. Also, depending on methodology used certain factors must be accounted for within the measurements themselves.

You could probably include yourself in the "credible third party measurements". Or do you blush easily?
I do and will always try to provide quality measurements, but it is not my place to call myself credible. After all, I am pretty sure I cannot maintain objectivity on a topic as awesome as me! :rolleyes::R:hide:
 

·
Registered
Joined
·
373 Posts
I understand the issue of bias and have read that particular Toole paper. It is also fair to say that the knowledge that bias exists will mitigate that bias to some extent.

If you do not believe that higher price always equates to better quality, then you are going to listen more critically for evidence of quality and are also likely to arm yourself with the tools required to do that critical listening.

Critical thinking/reasoning is a skill that anyone can acquire and apply to decision making to reduce bias.
 

·
Registered
Joined
·
1,249 Posts
Critical thinking/reasoning is a skill that anyone can acquire and apply to decision making to reduce bias.
I meet very few people who can actually carry through a logicaly reasoned appraisle through to final decision with it being tainted by either saving a dollar or owning some form of status symbol.

One of the things I enjoy most about not reading hifi magazines is that when I walk into a shop, I only recognise half of the brand names on the floor. One time I dismissed a $3000 pair of Dali's becasue their tonal quality was up the creek and the off axis response was shocking. It turns out they are the cheapest of the Dali range and were over priced on the ticket. For me this is the perfect example of what perceptual research is discovering.
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #17
drf makes a great point here: No one can escape bias regardless of knowledge level, it just isn't possible considering the human condition. Of course, conducting listening tests in blinding conditions is not a reasonable proposition the majority of the time because of this why I recommend and personally use a linear, non-resonant reference pair of headphones for comparison. While far from ideal such an approach at least allows one to make direct comparison between a known set of standards and another speaker.
 

·
Registered
Joined
·
373 Posts
I meet very few people who can actually carry through a logicaly reasoned appraisle through to final decision with it being tainted by either saving a dollar or owning some form of status symbol.
I drive a Ford Escort. There is no way I'm going to be biased by owning a status symbol. ;) :R

I agree, but that does not mean it can't be done. It is a learned skill.

Andrew, if we follow your reasoning to its logical end, we come to the conclusion that the only thing to do is to ask an expert what to buy.

1. we can't hope to eliminate bias no matter what.
2. we can learn to listen properly, but it is very very difficult and requires 'deep review'/learning.
3. The only reasonable listening conditions to eliminate bias are blind listening: an impossible situation for most of us.

Its the classical experts argument, all very logical and reasonable on the surface, but ...

Balancing that is the 80:20 rule. You can get 80% of the way there with reasonable effort. Bias can be reasonably limited by concious effort and the proper application of critical thinking.

I would put it to you that believing that bias cannot be reasonably limited is, um, biased. ;)
 

·
Registered
Joined
·
373 Posts
That headphones lisening thing is a good idea. I am slowly coming to the conlcusion that I am only an audiophile wanna-be though. I would have a hard time buying an expensive pair of headphones just to evaluate speakers. :)
 

·
Banned
Joined
·
1,131 Posts
Discussion Starter #20
1. we can't hope to eliminate bias no matter what.
True, even blinded tests can be easily confounded if proper steps are not taken.

2. we can learn to listen properly, but it is very very difficult and requires 'deep review'/learning.
I never said or implied this, rather correlating measurement to listener preference is what requires a large investment of time and effort.

Listening for quality itself is an easily learned trait as shown by Olive's study comparing trained and untrained listeners. The real problem is other confounds: short auditory memory, brand bias, loudspeaker room interaction etc...

Additionally, the lack of any reference to what is truly high quality makes such judgments harder for the 'uninitiated'. In fact, this last statement can likely be extended to those who have extensive listening experience in one major regard: panel resonance. This anomaly is so common in virtually all loudspeakers, regardless of price, it seems to be an accepted trait by most despite its correlation with poor listener ratings.

The only reasonable listening conditions to eliminate bias are blind listening: an impossible situation for most of us.
This is exactly why I recommended using a high quality linear pair of headphones that has no audible resonances. Such a unit at least allows for a reference to quality when auditioning, but certainly does not remove bias.


Balancing that is the 80:20 rule. You can get 80% of the way there with reasonable effort. Bias can be reasonably limited by concious effort and the proper application of critical thinking.


I would put it to you that believing that bias cannot be reasonably limited is, um, biased. ;)
No one is immune to bias, it just isn't possible. Certainly bias can be minimized, but each person will have their own definition of reasonably limited. Clearly this is where we differ. If ones goal is to have a good time and audition speakers certainly blinded comparison is not needed, but if the goal is ideal fidelity even minute amounts of bias can effect the outcome. Couple this with loudspeaker room interaction introducing yet another massive confound and it is clear that simple listening tests can find a speaker one likes, but not ideal. The end point being: listening tests are not a feasible option most of the time if this is ideal fidelity is the goal, but with proper correlation measurement to preference the ideal speaker for a specific price class is chosen.
 
1 - 20 of 27 Posts
Top