I think DBX or even sophisticated Dolby can't suppress certain harmonics selectively, only bands. Of course, like vinyl records and tube amps, suppressing the higher frequencies just might sound better.
You are right to mention that HD doesn't correct for loudness contour. But then it is rare to see that correction applied in other kinds of assessments.
As someone who went down the rabbit hole studying airport noise long ago, there is just no single-number approach to characterizing annoyance of that sort. A few metrics are needed such as loudness, duration, repetition, time of day, character of sound, etc. Good thing audio systems aren't as multi-dimensional as airport communities.
B.
Well said, sort of. But it is important to understand the difference between erroneous outputs and tests. Think of the tests like the seven blind men parable, each (initially) grabbing some different part of an elephant.
The tests don't define the DUT's "shape" although they can be read to provide some insight into the source of the error like poor polar response.... if you are using a test that can indicate that.
The general object here is to find the fewest tests that provide the best evaluation of "the elephant".
B.
With feedback I think the THD is less relevant than ever before.
You can take any system and reduce both THD and IMD with feedback at the price of a brick wall THD rise, and phase problems.
If you know how much local and global feedback is used in a system or open loop you can better judge the THD.
If THD was the only factor all amps would have a lot of gain and feedback.
With loudspeakers, THD is very important, as well as how dynamic the driver is. Standing waves too is difficult to judge from the impulse response and waterfall patterns.
You can take any system and reduce both THD and IMD with feedback at the price of a brick wall THD rise, and phase problems.
If you know how much local and global feedback is used in a system or open loop you can better judge the THD.
If THD was the only factor all amps would have a lot of gain and feedback.
With loudspeakers, THD is very important, as well as how dynamic the driver is. Standing waves too is difficult to judge from the impulse response and waterfall patterns.
An objective test will never exist because speakers are used in different environment for different purposes, and many only sell because of marketing or bad sound, or cool effect.
Until then an educated person can see which speaker is good with the click of a mouse and a little computer simulations from the data available from sources such as:
audioXpress | audioXpress Magazine. Advancing the Evolution of Audio Technology - Since 1970
HiFiCompass - всё для акустических систем и не только |
Driver manufacturers specs sheets.
SoundStageNetwork.com | SoundStage.com - Loudspeakers
Humble Homemade Hifi
DIY Loudspeaker Projects Troels Gravesen
Many loudspeakers databases and speakers sites with detail driver specs.
http://www.zaphaudio.com
You have to see what is important for you and your budget, info is available.
Until then an educated person can see which speaker is good with the click of a mouse and a little computer simulations from the data available from sources such as:
audioXpress | audioXpress Magazine. Advancing the Evolution of Audio Technology - Since 1970
HiFiCompass - всё для акустических систем и не только |
Driver manufacturers specs sheets.
SoundStageNetwork.com | SoundStage.com - Loudspeakers
Humble Homemade Hifi
DIY Loudspeaker Projects Troels Gravesen
Many loudspeakers databases and speakers sites with detail driver specs.
http://www.zaphaudio.com
You have to see what is important for you and your budget, info is available.
Au Contraire
Just for example:
JBL Harmon have a complete set of objective tests. If a speaker passes the objective test protocol the speaker will pass the preferred subjective listening test as well.
A complete linkage between objective laboratory tests and subjective listening tests.
Thanks DT
well harmon has its own tests, so they are not objective, they only apply to their particular subjective listening objective. I never heard of such a test... except quality control with a reference then they pass each unit within a normal deviation...
you have to choose what you want the driver to do, then you look at the specifications and make a decision, so there is no objective tests.
EX a driver can be xo at many points, depending on the design, so, in that context, an 'objectively' superior driver could be inferior in one area on particular, an the inferior driver restrained to that particular range or power usage or distance could be superior.
A good example is line array speakers, they use smaller driver which are not suitable for 2 - 4 ways, but their number accumulate and in that use they are superior.
you have to choose what you want the driver to do, then you look at the specifications and make a decision, so there is no objective tests.
EX a driver can be xo at many points, depending on the design, so, in that context, an 'objectively' superior driver could be inferior in one area on particular, an the inferior driver restrained to that particular range or power usage or distance could be superior.
A good example is line array speakers, they use smaller driver which are not suitable for 2 - 4 ways, but their number accumulate and in that use they are superior.
I've been reading a great book, "AIQ" about artificial intelligence. For example Netflick's suggestions for you (OK, used to be called Operations Research or Bayesian stats).
What Toole made for Harmon is a set of tests that point to the quality dimension. (OK, there is a difference between having "a test" like polar response and knowing "the quantitative assessment of test results" once you have the polar plot in your hands.
There is some approach mid-way between the Netflix Recommender AI and simple THD that gives us the least mean squares prediction of sound quality.
B.
What Toole made for Harmon is a set of tests that point to the quality dimension. (OK, there is a difference between having "a test" like polar response and knowing "the quantitative assessment of test results" once you have the polar plot in your hands.
There is some approach mid-way between the Netflix Recommender AI and simple THD that gives us the least mean squares prediction of sound quality.
B.
I think those tests points to the wrong dimension. With a computer calculated best audible quality factor you end up making mediocre speakers that all sound the same.
and why do you need 'objective' driver specs when you know that the best tweeters technology , thd, response, were achieved in the 1990s, there are no improvements in tweeter technology since then.
for woofers, beside harbeth injected poly cones, the best alinco, paper cones were designed in the 1980s there is not much changes since then, those drivers are quite easy to find by research, makes a bit pointless any comparison, their THD all all harmonics are lower than all other drivers and they have smooth response.
and why do you need 'objective' driver specs when you know that the best tweeters technology , thd, response, were achieved in the 1990s, there are no improvements in tweeter technology since then.
for woofers, beside harbeth injected poly cones, the best alinco, paper cones were designed in the 1980s there is not much changes since then, those drivers are quite easy to find by research, makes a bit pointless any comparison, their THD all all harmonics are lower than all other drivers and they have smooth response.
It is time to listen to the music.
Hello,
This thread needs more posts.
This distortion thing should be broken into manageable pieces.
1 sources
2 Line level devices
3 Amplifiers
4 Speakers
For example get the amplifier to pass some basic tests like THD + N less than some number, IM less than some number and then the other screechy bad sounds will also fall into line. Then let the golden ear folks argue over if anything else is important to measure.
Then move on to measuring the speakers and also do some basic Pass Fail tests.
Pass all the basic tests and you are 99% there.
I have spent far more on test equipment than audio equipment that I can listen to.
It is time to listen to the music.
Thanks DT
Hello,
This thread needs more posts.
This distortion thing should be broken into manageable pieces.
1 sources
2 Line level devices
3 Amplifiers
4 Speakers
For example get the amplifier to pass some basic tests like THD + N less than some number, IM less than some number and then the other screechy bad sounds will also fall into line. Then let the golden ear folks argue over if anything else is important to measure.
Then move on to measuring the speakers and also do some basic Pass Fail tests.
Pass all the basic tests and you are 99% there.
I have spent far more on test equipment than audio equipment that I can listen to.
It is time to listen to the music.
Thanks DT
Geddes and Lee studied distortion of loudspeakers. Original papers were published 2001-2003 - here GedLee LLC
Another thread about this The Gedlee Metric Demystified
Another thread about this The Gedlee Metric Demystified
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.