• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

How we could finally pin down flowery audiophile subjective descriptions

kemmler3D

Major Contributor
Forum Donor
Joined
Aug 25, 2022
Messages
3,008
Likes
5,608
Location
San Francisco
Many of us throw up our hands when equipment is described as "fast", "slow", "crisp", "warm", etc. It seems impossible to relate these terms to measurable characteristics.

I have a slightly more optimistic view, in that subjective descriptions must be correlated with what people hear, and what people hear tends to be correlated (imperfectly) to measurable output.

To translate subjective descriptions into objective measurements (or the other way around, which might be more interesting), I propose that a machine learning model could be used.

The model would correlate subjective terms used (how often people say a speaker is "crisp", for example) with the measurements of the equipment relative to the median measurements of all equipment in its category.

The output would be a map of how semantically close certain audiophile words are to each other, and how use of those words correlates with measurable characteristics. Imagine a word cloud that groups words like "sharp, bright, tinny" together, and then displays measured characteristics that correlate with those terms - elevated response above 4khz, above-average H3 distortion in that range, etc.

This would be interesting in its own right, but if the ML was sophisticated enough, and fed enough data, it might also reveal trends in preference / subjective experience that go beyond the current preference score models. For example, you might unexpectedly find that some aspect of vertical directivity correlates with "warmth" or "speed", for example. I don't know.

I don't have anywhere near the skills to execute such a project, but it seems like a way to solve the "problem" of people using flowery language that, to many of us, is currently worse than useless. It might also reveal that things audiophiles consider to be "beyond science" are actually very well correlated with simple measurements. Which would be progress in and of itself.
 

fpitas

Master Contributor
Forum Donor
Joined
Jul 7, 2022
Messages
9,885
Likes
14,191
Location
Northern Virginia, USA
Problem is, the subjective sorts for the most part won't reach for the microphone and nail down their impression.
 
OP
kemmler3D

kemmler3D

Major Contributor
Forum Donor
Joined
Aug 25, 2022
Messages
3,008
Likes
5,608
Location
San Francisco
Problem is, the subjective sorts for the most part won't reach for the microphone and nail down their impression.
Whenever the measurements for a given speaker are known, you can then go and find any subjective description of the same speaker to correlate with the measurements. It wouldn't need to happen in the same review. As long as are confident that any given person is listening to the same speaker, it would be a valid input.

Given that the room has so much influence, we'd probably want dozens of impressions per speaker, to average out the effect of the room on subjective reactions, but as long as you have enough of these, it should work.

Overall you'd want dozens of reactions per speaker, and probably 50-100+ speakers fed into the model, to get usable results.

The strength of this approach is you don't need to do actual tests in person. The weakness is you need a ton of data to get statistically useful results.
 

Speedskater

Major Contributor
Joined
Mar 5, 2016
Messages
1,610
Likes
1,333
Location
Cleveland, Ohio USA
J. Gordon Holt (RIP) published 'The Audio Glossary' in 1990.
Now available at Stereophile:

But audiophile bloggers will find new ways or meanings with words.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
One word................plankton...............................SBAF terminology.
 
OP
kemmler3D

kemmler3D

Major Contributor
Forum Donor
Joined
Aug 25, 2022
Messages
3,008
Likes
5,608
Location
San Francisco
Devil's advocate here: I take 'plankton' to mean 'very fine detail in the recording'. Perception of "plankton" would therefore depend on a good transient response and very low noise and distortion, particularly IMD. So I think you can relate even something like "plankton" to measurable quantities. The difficulty comes in knowing whether anyone interprets a colorful term the same way.
 
OP
kemmler3D

kemmler3D

Major Contributor
Forum Donor
Joined
Aug 25, 2022
Messages
3,008
Likes
5,608
Location
San Francisco
Two problems:

1) Expectation bias. There is no way to get around expectation bias without performing a test (or audition) blind. None. The power of expectation bias in regards to sighted listening is not only powerful, but (I believe) vastly underrated in its effect.




2) I don't believe that's true. Different listening distances and different listening axes are, I am convinced, accountable for many otherwise well-meaning assessments of speaker "sound". In particular, I have seen (and heard) stand-mount speakers in other people's homes that are set up off-axis, and the result was noticeably different than the same speaker ten degrees higher (or lower). Some people don't understand the existence (much less the effect) of a "correct" listening axis, and others have the installation governed by appearance only (WAF).

I'm not saying that this is true for all speakers, and I certainly don't know what percentage of speakers (or installations) display this characteristic. But I do know that it exists, and my guess (and it's only a guess) is that the percentage is high enough to invalidate a good percentage of the results.

I know, I know .... lots of weasel words in there ("some", "good", "many"). Sorry about that.. :facepalm:

Jim
Valid points, the more variability in how a speaker might sound to someone (including due to room effects, listening angle, etc.) the more impressions you need to find the "true average", i.e. the underlying subjective qualities of the speaker. Statistically, you could just call this a noisy signal, which in theory can be overcome by more samples.

The theory is that if you average out enough impressions, they will converge on a set of terms, perhaps one set per listening environment, but you get the idea.

I tend to agree that there are very few speakers with enough impressions out there to overcome the 'noise', but it's a thought.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,198
Likes
16,981
Location
Riverview FL
One word................plankton...............................SBAF terminology.

Yeah!

1668212890796.png


Translucent!

Black Background!

Plus. it's scientifical.
 
OP
kemmler3D

kemmler3D

Major Contributor
Forum Donor
Joined
Aug 25, 2022
Messages
3,008
Likes
5,608
Location
San Francisco
I mostly agree with these definitions / charts... it's not that hard to come up with one yourself if you have a mind to. My thought was more that we could pin down people who don't even have this degree of connection to the actual sound.

Interesting that they differ by +/- 100hz here and there for the same terms. Even people who feel they have a strong grip on this stuff don't perfectly agree. Not a huge surprise but kinda illustrates the point, too. These terms rely on loose averages and not exact numbers.
 

Martin Takamine

Active Member
Joined
Sep 8, 2019
Messages
270
Likes
628
Location
East Coast
Until subjective reviewers post their audiogram results from certified technicians that administer tests using calibrated audiology equipment then their review is useless. The reasons being is I don't know their range of hearing or what frequencies they can't hear within their range.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
Until subjective reviewers post their audiogram results from certified technicians that administer tests using calibrated audiology equipment then their review is useless. The reasons being is I don't know their range of hearing or what frequencies they can't hear within their range.
You don't trust that they can hear if they don't hear so well?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Many of us throw up our hands when equipment is described as "fast", "slow", "crisp", "warm", etc. It seems impossible to relate these terms to measurable characteristics.

I have a slightly more optimistic view, in that subjective descriptions must be correlated with what people hear, and what people hear tends to be correlated (imperfectly) to measurable output.

To translate subjective descriptions into objective measurements (or the other way around, which might be more interesting), I propose that a machine learning model could be used.

The model would correlate subjective terms used (how often people say a speaker is "crisp", for example) with the measurements of the equipment relative to the median measurements of all equipment in its category.

The output would be a map of how semantically close certain audiophile words are to each other, and how use of those words correlates with measurable characteristics. Imagine a word cloud that groups words like "sharp, bright, tinny" together, and then displays measured characteristics that correlate with those terms - elevated response above 4khz, above-average H3 distortion in that range, etc.

This would be interesting in its own right, but if the ML was sophisticated enough, and fed enough data, it might also reveal trends in preference / subjective experience that go beyond the current preference score models. For example, you might unexpectedly find that some aspect of vertical directivity correlates with "warmth" or "speed", for example. I don't know.

I don't have anywhere near the skills to execute such a project, but it seems like a way to solve the "problem" of people using flowery language that, to many of us, is currently worse than useless. It might also reveal that things audiophiles consider to be "beyond science" are actually very well correlated with simple measurements. Which would be progress in and of itself.

How to deal with flowery words like 'syrupy', 'chocolaty' etc.
And how to deal with words that can have many meanings for different people. For instance 'analog' sound or 'digital' sound, glare, punch, oomph for instance.
Most descriptions are used by people that think what a description someone mentioned meant and use that erroneously or in a (non technical) way closer to their own line of thoughts, experience.

A 'conversion' from 'flowery' wording by various individuals may differ from person to person and use different 'vocabulary' than those of others, or even the same words but meaning something different for that person. That 'flowery wording' conversion to 'standardized technical' wording/properties would have to take this under consideration.

Maybe it would be best to have an 'agreed upon' list of descriptors with a clear description for each word with what is meant by it and that people try to adhere to or are pointed to that list.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
TAKE MEASUREMENTS.

chicken - egg problem. To make measurements have any meaning to individuals with their own 'vocabulary' they need to know how to convert squiglies to their vocabulary (understand a full measurement suite) and that takes quite a bit of knowledge, experience. Even if 'agreed upon' descriptors are used (as an explanation along with the measurements) this might not be fully understood by people with a very different vocabulary. They would have to use the elaborate 'descriptors descriptions' list to understand what the 'agreed upon' descriptors means.

For ASR, for instance, such a vocabulary could be agreed upon.... maybe.
 

raif71

Major Contributor
Joined
Sep 7, 2019
Messages
2,333
Likes
2,535
Many of us throw up our hands when equipment is described as "fast", "slow", "crisp", "warm", etc. It seems impossible to relate these terms to measurable characteristics.

I have a slightly more optimistic view, in that subjective descriptions must be correlated with what people hear, and what people hear tends to be correlated (imperfectly) to measurable output.

To translate subjective descriptions into objective measurements (or the other way around, which might be more interesting), I propose that a machine learning model could be used.

The model would correlate subjective terms used (how often people say a speaker is "crisp", for example) with the measurements of the equipment relative to the median measurements of all equipment in its category.

The output would be a map of how semantically close certain audiophile words are to each other, and how use of those words correlates with measurable characteristics. Imagine a word cloud that groups words like "sharp, bright, tinny" together, and then displays measured characteristics that correlate with those terms - elevated response above 4khz, above-average H3 distortion in that range, etc.

This would be interesting in its own right, but if the ML was sophisticated enough, and fed enough data, it might also reveal trends in preference / subjective experience that go beyond the current preference score models. For example, you might unexpectedly find that some aspect of vertical directivity correlates with "warmth" or "speed", for example. I don't know.

I don't have anywhere near the skills to execute such a project, but it seems like a way to solve the "problem" of people using flowery language that, to many of us, is currently worse than useless. It might also reveal that things audiophiles consider to be "beyond science" are actually very well correlated with simple measurements. Which would be progress in and of itself.
I like that some of us will depend on this AI machine to interpret audio data into flowery words like we're putting all our "hopes" on the APx555. :)
 

Axo1989

Major Contributor
Joined
Jan 9, 2022
Messages
2,804
Likes
2,805
Location
Sydney
Devil's advocate here: I take 'plankton' to mean 'very fine detail in the recording'. Perception of "plankton" would therefore depend on a good transient response and very low noise and distortion, particularly IMD. So I think you can relate even something like "plankton" to measurable quantities. The difficulty comes in knowing whether anyone interprets a colorful term the same way.

Plankton is a great analogy tbh.


I really like a lot of music that sounds like that.

This would be interesting in its own right, but if the ML was sophisticated enough, and fed enough data, it might also reveal trends in preference / subjective experience that go beyond the current preference score models. For example, you might unexpectedly find that some aspect of vertical directivity correlates with "warmth" or "speed", for example. I don't know.

That would be the fun bit.

Personally I think the problem is less when people use words, more when they use them as meaningless cliche.
 

Ricardus

Addicted to Fun and Learning
Joined
Mar 15, 2022
Messages
843
Likes
1,153
Location
Northern GA
chicken - egg problem. To make measurements have any meaning to individuals with their own 'vocabulary' they need to know how to convert squiglies to their vocabulary (understand a full measurement suite) and that takes quite a bit of knowledge, experience. Even if 'agreed upon' descriptors are used (as an explanation along with the measurements) this might not be fully understood by people with a very different vocabulary. They would have to use the elaborate 'descriptors descriptions' list to understand what the 'agreed upon' descriptors means.

For ASR, for instance, such a vocabulary could be agreed upon.... maybe.
It's even worse than the chicken egg thing.

The point is, the measurements give us the vocabulary we need. Numbers. DONE.
 
Top Bottom