[gevorg]

It is very difficult to describe sound quality, let alone quantify it, since everyone has their own preferences such as EQ, coloration, etc. Lets start with the baseline pro audio industry standard, the "neutral" or "monitor" sound, which is the preference of many audiophiles too. That is, to recreate the original recording with little variations as possible.

Modern measuring equipment can tell us quite a lot about sound properties such as frequency response and dynamic range, but its still not enough to give the final word on the overall sound quality. This means we have to resort to subjective evaluation of sound quality, which of course varies from one person to another. We can still use the measurements too if we combine them with subjective evaluation by some weighted-average formula. The weighted-average formula can even put some weight to other information in determining sound quality, such as price and product release date.

Until technology advances to be able to accurately measure sound quality, it seems the only way to do it would be by applying statistics to subjective listening results. If the results are statistically significant and contain quality data, it should be possible to get a good sense of sound quality of a given equipment. The results would be even more accurate if the end-user will also be able to specify their preferences, for example: ignore price.

To use statistics there are two major factors to consider: point of reference when evaluating equipment, and what statistical distribution to use to define sound quality (normal, beta, etc).

What do you think is a feasible method to quantify sound quality?

 

[royalcrown]

The whole notion of measuring sound quality, let alone any sort of quality, is nonsensical. Quality is by its nature qualitative and subjective, and since it is subjective it will necessarily vary from person to person.

Measurements are used to assess accuracy, not quality, much in the same way as color analysis can quantify the accuracy of a reproduction of a painting. Some people might prefer a print of a portrait with a filter applied to it (and most don't care beyond a point), but art collectors and aficionados will want the most accurate print possible to distill, as far as possible, only the original artwork's aesthetics. Likewise, some people prefer colored music, preferring to tweak the sound as they see fit. Most don't care and use earbuds. People that are interested in hi-fi (in the very strict sense of the word), among others including those involved with audio production, want the most accurate playback possible to distill, as far as possible, the original recording's sound. No approach is better than the others.

 

[haloxt]

gevorg, gathering data that way would cost a lot of money. It'd be cheaper to just keep everyone supplied with alcohol so they think 128kbps sounds good.

If some software comes about that can read people's facial expressions and tell with uncanny accuracy how they are feeling, that could make for a pretty hilarious approach to gauging sound quality even though it won't necessarily deal with the major issue of personal preferences.

Before we can quantify sound quality we'd have to understand sound perception to a great degree, and who knows, figuring out how to create virtually perfect sound quality may come before figuring out how to quantify sound quality. And who cares whether the chicken or the egg came first when you can buy a nice chicken ready to eat at the supermarket.

 

[PhilS]

Quote:

Originally Posted by royalcrown /img/forum/go_quote.gif The whole notion of measuring sound quality, let alone any sort of quality, is nonsensical. Quality is by its nature qualitative and subjective, and since it is subjective it will necessarily vary from person to person.

I've often had the same thought when people talk about, for example, a 30% improvement in sound quality from a change they made to their system, whether it be a new headphone, amp, or whatever. How does one quantify what a 30% improvement is such that it can have any meaning to anyone else?

For similar reasons, I'm not sure it is always accurate to apply the notion of "diminishing returns" to improvements in sound quality. If sound quality is subjective, and impossible to measure in terms of applying a standard, it's somewhat questionable to say that an increase in expenditure must bear some sort of linear relationship with sound quality. Sometimes it might, and sometimes it might not.

 

[ramenmeal]

i'd say sound quality is really just trying to recreate the original sound. With acoustic instruments, thats easy to define. Seems like you could then measure sound quality by measuring the difference between the original and the recreated. no?

 

[anwaypasible]

everything starts with a flat frequency response coming from the speakers.
doesnt mean what colour is being produced compared to the frequency response being flat.

dynamics of speakers are just the number of things/notes/soundwaves that the driver can produce at once.

the economy will try to sell you dynamics as premium.. despite a flat frequency response.
and if you try to flatten that frequency response with an EQ, the dynamics go down.
the speaker says, okay.. if you want me to play these notes louder and these notes softer, i cant do anything else - theres no room in here to do so.
if you dont touch the equalizer, then the speakers say.. hey, i got some free time to do more than one note/tone/sound at once.

kinda like a computer and the cpu load.
when the cpu load is at 100% there isnt much else that can be processed at that moment.
so a computer processor that can multi-task MORE wont go up to 100%

sometimes adjusting a driver with an EQ is like opening up 6 or 7 large programs at once, therefore there is no processor left to do anything.
big things like having a resonant frequency far or close away from the drivers impedance peak/lowest point can have a substantial difference on whether a driver will efficiently accept an EQ or not.
if the resonant frequency is close to the highest impedance point, the speaker may see the EQ adjustments as 'something lite/easy' and will only go up in load 15%
while a speaker with a resonant frequency very low and an impedance peak very high, would make that speaker have a headache if you tried to flatten its frequency response.

an EQ adjustment based on the impedance curve might leave your speakers sounding coloured and detailed, while an EQ adjustment based on frequency response might leave your speakers sounding like cardboard.

there are 2 parts to the speakers soul & personality. the derivitives consist of impedance curve and magnet/coil characteristics that equal the frequency response.

and the impedance curve is where all the fast-paced sensitivity happens.
it is like the drivers energy/hyperactivity level where nanoseconds and milliseconds count.
if you tailor your equalizer towards the frequency response and not the impedance curve, you will notice that the speaker has to work hard to create the adjustment you want and thus you hear the latency in the driver that makes the speaker sound like cardboard (no dynamics)

what is 'standard' is fed to the entire planet by the most respectable people in audio.
generally we simply tend to agree, and thus the standards come easy.
other producers can 'allow' a decision and go home to their 'preference'
those producers will have a legacy that stays with them.. a 'known & loved for this' if you will.

 

[Uncle Erik]

Don't forget the physiological differences. There are differences in ear shape and other physical attributes that affect the way we hear.

Not to mention the psychological differences. People develop a baseline for what they perceive as "right" depending on a lifetime of exposure to music. Sadly, many people have very limited exposure to live music. If 99% of your listening has been to crappy car stereos, cheapie HTIB systems, etc., then that will skew your perceptions. People who have listened to bloated bass and attenuated highs for years generally do not like equipment with tight bass and a genuine top end. They'll call something like that "bass light" and complain about overemphasized highs no matter how flat something measures.

There are probably too many variables to quantify the sound of a piece of equipment in a meaningful way for people.

 

[astroid]

I have always found this topic interesting, i get a lot of Hifi mags and frequently become annoyed and confused with the reviews.
What does an AMP sound like, its silent until you feed it a signal, the question is does it colour that signal in any way (and has the source colored it already!).
To find that out you need to know what a source would sound like uncolored, but CD players dont produce sound , amps do!!!!
Is it the cable that is making it sound muddy, or the power supply , or the amp or is it the recording????
The mags generally swap out the subject for other examples and note the differences but which one is 'right'?
Total minefield...

 

[echelon]

Quote:

Originally Posted by gevorg /img/forum/go_quote.gif What do you think is a feasible method to quantify sound quality?

1) Play a set of sound waves in known frequencies into an amp.
2) Record the output via lineout, and then compare to the input.

You can now quantify how the amp has modified the original sound. If you do this test with 2 sets of cables, you can quantify whether different cables have different sound signatures.

Seems pretty simple; what am I missing?

 

[PhilS]

Quote:

Originally Posted by echelon /img/forum/go_quote.gif 1) Play a set of sound waves in known frequencies into an amp. 2) Record the output via lineout, and then compare to the input. You can now quantify how the amp has modified the original sound. If you do this test with 2 sets of cables, you can quantify whether different cables have different sound signatures. Seems pretty simple; what am I missing?

The OP is talking about quantifying or measuring sound quality.

 

[xnor]

Quote:

Originally Posted by PhilS /img/forum/go_quote.gif The OP is talking about quantifying or measuring sound quality.

Well I would define quality as the ratio between input and output, i.e. 'flat' or 'monitor' sound.
Why?
a) Because if the output is colored, EQ'd etc. it's very hard to undo that.
b) You can go anywhere from a flat response using EQ, dsp ...

@echelon: Though, you'd have to make sure that speakers are connected to it during the measurements.

 

[Slash47]

Can you quantify quality? It doesn't work for anything I know of... (Houses, cars, clothes, even tv's or computers, not to mention living things.)

Logic (science) only estimates, nothing more. Quality is mostly psychological or spiritual. In that sense, I think you could say quantification is already quite advanced with words like 'imaging' and 'warm' or 'micro dynamics' and I think numbers would be a downgrade, but that's a weird way of putting it, I know.

But I wouldn't disagree that the best amp and source would only have accurate analogue conversion, gain and nothing more, so you could say that distortion is the main quantity you'd qualify. But then again those numbers are so far down that you can hardly seem to make much out of 'm. Room acoustics and stuff like that creates more distortion when the sound finally reaches the ear, I'm sure. That's also a major screw-up with a lot of audiophile science, they measure at the wrong place. Who cares what comes out of your amp? Maybe a certain color makes the final product much more neutral. The same way the Acropolis in Greece is completely crooked but looks perfect when you look at it from Athens and there's blue in washing powder for white laundry. Tube amps might be doing something right in that department, but I have no idea what it is.

And I think that we should blame recordings even more often. I think amplifiers and speakers are a lot 'better' then we give them credit for.

 

[royalcrown]

Quote:

Originally Posted by PhilS /img/forum/go_quote.gif For similar reasons, I'm not sure it is always accurate to apply the notion of "diminishing returns" to improvements in sound quality. If sound quality is subjective, and impossible to measure in terms of applying a standard, it's somewhat questionable to say that an increase in expenditure must bear some sort of linear relationship with sound quality. Sometimes it might, and sometimes it might not.

I don't know if I'd go that far. The concept of diminishing returns is based in a trend, and I think it's a pretty good rule of thumb as far as audio goes. For the majority of people, with the majority of equipment, the trend works pretty well. Just because sound quality is subjective, that doesn't mean that there are similarities within those subjective experiences - even though there's an odd emu every so often that finds a gigantic improvement going from a 1,000 dollar component to a 2,000 dollar component, most people won't. From that point of view, I'd say it's better to play it safe and save your cash rather than dropping large sums of money hoping to hear a difference when there's a low chance that you will actually hear a difference.

 

[PhilS]

Quote:

Originally Posted by xnor /img/forum/go_quote.gif Well I would define quality as the ratio between input and output, i.e. 'flat' or 'monitor' sound.

Well, you can always define away the real issue.

 

[PhilS]

Quote:

Originally Posted by royalcrown /img/forum/go_quote.gif I don't know if I'd go that far. The concept of diminishing returns is based in a trend, and I think it's a pretty good rule of thumb as far as audio goes. For the majority of people, with the majority of equipment, the trend works pretty well. Just because sound quality is subjective, that doesn't mean that there are similarities within those subjective experiences - even though there's an odd emu every so often that finds a gigantic improvement going from a 1,000 dollar component to a 2,000 dollar component, most people won't. From that point of view, I'd say it's better to play it safe and save your cash rather than dropping large sums of money hoping to hear a difference when there's a low chance that you will actually hear a difference.

My point is illustrated by this example. Suppose I hate sibilance. I have a system that costs $1000, and it sounds pretty good, but still has some sibilance. I upgrade the system by spending $500 on something. This essentially eliminates the sibilance. The marginal cost was of the improvement was 50%. What's the marginal improvement in sound quality? How to you measure it? The marginal "value" of the improvement to the listener might be 10%, 100%, or even 1000%.

Sure, the concept of diminishing returns is generally valid in this area. But my point is that one has to be careful about saying it always applies or even often applies, because there is no standard by which we can meaningfully measure what an improvement in sound quality means to a particular listener.