Ideas for more meaningful speaker measurements

[Holmz]

Reflections aren't a bad thing if the goal is to emulate a bigger/better room. See the combing in a real room.
DSP takes care of the perceived balance. As long as nothing sticks out this is quite doable.

I am not saying that the room is bad.
(It is better than living rough in the rain or snow, or desert.)

Either we care about a direct path, or we care about reflections.
In the time domain, and correlation space, one can separate the direct path from the reflected path.

Are you saying that we want the direct sound altered, in order to account for the reflections?

 

[Wesayso]

@Holmz I do think you get the wrong impression of what it is I do or have. I absorb early reflections, as seen from my listening position. That means i stole a lot of energy from the room. I simply don't have the space to redirect it, which would have been a better choice. So I'm using ambient speakers, placed behind the listening position and aimed outwards (away from the LP) to reflect off of nearby objects. It is those speakers that receive a decorrelated mix. The direct sound is not altered in any way, shape or form (*). That original sound still gets send to the main speakers. The ambient speakers give clues of a fake, but better room to the listener, mostly arriving from lateral angles. These cues come in after about 17 to 20 ms after the direct sound, much like a Haas kicker was used or a live end/dead end room.
I'm not saying anyone should want to do anything I do. I just prefer this route, as I get more pleasant imaging (due to having a Reflection Free Zone) while still experiencing envelopment and ambience. It is a mix that adapts to the song being played, so not the same sauce on everything.

The idea stems from quite a while ago with some experiments done in an even smaller space, a car. Back then a member of a different (car audio) forum triggered the interest of what could be done with a "faked" back drop. I don't mind what others think or feel about it, as I'm in it for MyFi, not specifically HiFi in it's strictest sense as some put it. I like an enjoyable presentation, one that makes me feel I'm right there, at the take/recording itself. It all has to do with one's preference. I prefer having that RFZ (Reflection Free Zone) but need to make up for the energy lost/absorbed.
Much more info can be found in a long thread. Some reviews from friendly members that visited me can be found there too. Links are present in the first post to make it somewhat easy to navigate. See it as my own version of experiments Siegfried Linkwitz used to do (may he RIP). Same idea basically, different way of going about it.

On with the subject at hand, sorry for this detour, even though parts of it may spark some interest for a different way of taking on the problems related to what's discussed in this thread. As said; the room will dominate a lot that's heard at the listening position. So I treat my system, room + speakers, as the query to solve. That's why I picked these specific speakers and do what I do. It's all DIY, which makes me able to cater to my wishes.

(*) = relatively speaking. There's more going on not related to this discussion.

 

[storing]

I gave you an example. I cannot even comprehend science as a whole, how on earth I will try to explain it?

Well, trying to explain it is what you did making super general and hence unnuanced statements on how it's supposed to work. Now I'm pretty sure you know it's not like that, or not, I don't really care, but the problem with respect to the forum is there are people here who have much less of a clue than we do. So they might read such general statements and think "oh, ok I learned today that in science measurements come last" and take that as the sole, unnuanced, truth. Which is is not nice.

 

[changer]

I do miss more definitive parameters that describe the actual stereo perception. The symptomatic trace of this deficit is that besides the ubiquitous and in some regard agreeably telling sonograms, clues on how the speaker presents the stereo music at the listening place, as opposed to how it performs quality-indexed, are usually part of the subjective postscript of a review, if stereo plays a role at all! But directivity is not nearly enough to help understand the rendering of a pair of stereo speakers.

When I first listened to a pair of Elac Carina FS 247.4, I thought they were too far apart, yet the positioning complied to the stereo triangle. The center image was darker than left and right. The speakers I grew accustomed to, some waveguide two-way speakers with high directivity, have a very solid center image. You can spread them further apart than the ideal even-sided triangle, the center is still more even than with the Elacs. The Carina’s image is also behind the speakers. Some say this is a property of the crossover typology. If so, this should be a factor in reviews. My own speakers work on LR4 filters that is said project forward, and also the waveguides, reducing early reflections, tend to project the acoustic event happening towards the listener.

This is so important for the listening impression, it is like another world. To only represent this in the side notes and without more rigorous explanation leaves me a bit clueless. How could this elementary way how I listen to a recording be not more than a byline? Especially after so many speakers have been investigated with drivers playing separately, which enables to check the crossover typology/acoustic filter.

 

[sarumbear]

Well, trying to explain it is what you did making super general and hence unnuanced statements on how it's supposed to work. Now I'm pretty sure you know it's not like that, or not, I don't really care, but the problem with respect to the forum is there are people here who have much less of a clue than we do. So they might read such general statements and think "oh, ok I learned today that in science measurements come last" and take that as the sole, unnuanced, truth. Which is is not nice.

I had no intention to mislead anyone. I urge any of those folk to educate themselves so that they don’t get mislead by mistake and never, ever take a few sentence in a post as gospel truth. To me that is utter stupidity.

The fact is when talking about science often many things are correct at the same time. It’s the context that matters.

 

[Head_Unit]

1) Closeness of sample-to-sample matching; and 2) are the cabinets physically inert?

Seems to me offhand that cabinet resonances should show up in distortion measurements, no? I was thinking about this when reading Harman/JBL engineering manager was it Sean Olive saying that sorry John Atkinson putting an accelerometer on the cabinet is not meaningful you'd need to integrate the cabinet output. As for sample matching, that's always an issue. AVR XYZ measured 99 dB SINAD, so it must be better than AVR QRS which only measures 96 dB. OR is it really? Maybe one was built on a happy day and the other on the day the air conditioning and toilets all broke.

 

[Head_Unit]

Do we need the 32 tone test for loudspeakers? Probably could get a few people to do one with REW at the LP just to get an idea how much the results of such vary.

I really like the 32 tone test especially for the visual. When you say "vary" do you mean sample-to-sample? Or if different room effects? Or ???

What I would REALLY like to see would be coax tweeter responses with the woofer all the way out, neutral, all the way in. I'm really curious how much the frequency response is affected. This is not a standard test, more like a several-off.

 

[dshreter]

I do miss more definitive parameters that describe the actual stereo perception. The symptomatic trace of this deficit is that besides the ubiquitous and in some regard agreeably telling sonograms, clues on how the speaker presents the stereo music at the listening place, as opposed to how it performs quality-indexed, are usually part of the subjective postscript of a review, if stereo plays a role at all! But directivity is not nearly enough to help understand the rendering of a pair of stereo speakers.

When I first listened to a pair of Elac Carina FS 247.4, I thought they were too far apart, yet the positioning complied to the stereo triangle. The center image was darker than left and right. The speakers I grew accustomed to, some waveguide two-way speakers with high directivity, have a very solid center image. You can spread them further apart than the ideal even-sided triangle, the center is still more even than with the Elacs. The Carina’s image is also behind the speakers. Some say this is a property of the crossover typology. If so, this should be a factor in reviews. My own speakers work on LR4 filters that is said project forward, and also the waveguides, reducing early reflections, tend to project the acoustic event happening towards the listener.

This is so important for the listening impression, it is like another world. To only represent this in the side notes and without more rigorous explanation leaves me a bit clueless. How could this elementary way how I listen to a recording be not more than a byline? Especially after so many speakers have been investigated with drivers playing separately, which enables to check the crossover typology/acoustic filter.

I don't think there are definitive parameters to describe stereo perception. The measurements come first intentionally, because that is the audio science part of the review.

The subjective listening impressions are nice and can help relate what the measurements mean to the listening experience, but it's well known that uncontrolled subjective impressions are not reliable - so I believe they come last so as to not be perceived as superseding the measurements. Not to mention that a review of how speakers sound in Amir's house has limited applicability to other rooms.

 

[changer]

I don't think there are definitive parameters to describe stereo perception. The measurements come first intentionally, because that is the audio science part of the review.

The subjective listening impressions are nice and can help relate what the measurements mean to the listening experience, but it's well known that uncontrolled subjective impressions are not reliable - so I believe they come last so as to not be perceived as superseding the measurements. Not to mention that a review of how speakers sound in Amir's house has limited applicability to other rooms.

But it isn’t a subjective matter, it is a result of engineering. Please read again.

 

[Thomas_A]

What could be useful is to use existing spin data and make outputs for different speaker positions in stereo setups.

Second, to enable binaural codec filter from each loudspeaker that can be applied on music files and listened to by headphones.

 

[dshreter]

But it isn’t a subjective matter, it is a result of engineering. Please read again.

Ok, what are the definitive parameters then?

 

[Rick Sykora]

An interesting reveal when 2 Klippel systems test the same model speaker...

Edifier MR-4 review (by Erin)

Thanks to @hardisj for another review. This time it's the Edifier MR-4 (same model that Amir reviewed previously): Obviously, loudness limited. Use nearfield only. EDIT 2022-01-11: Video review is up Discuss!

www.audiosciencereview.com

 

[Rick Sykora]

I had no intention to mislead anyone. I urge any of those folk to educate themselves so that they don’t get mislead by mistake and never, ever take a few sentence in a post as gospel truth. To me that is utter stupidity.

The fact is when talking about science often many things are correct at the same time. It’s the context that matters.

Everyone, please cut the bantering and keep your posts on topic.

Thanks!

Rick

 

[dwkdnvr]

Do we need the 32 tone test for loudspeakers? Probably could get a few people to do one with REW at the LP just to get an idea how much the results of such vary.

Personally, I think this would be very interesting. I think you'd want to filter the amplitude of the components according to a 'music spectrum' or at least 'pink noise' curve rather than having them all at equal/full amplitude. I think it might well be very revealing, particularly in terms of things like the benefits of a 3-way vs a 2-way even if the 2-way has well controlled radiation.

 

[sarumbear]

Please cut the bantering and keep your posts on topic.

Thanks!

Rick

Good advice but it would be fairer if it’s not aimed at a single person.

 

[changer]

Ok, what are the definitive parameters then?

As I said already, different crossover filter functions are held responsible for shifiting the position of the phantom image back and forth.
The nature of the center image is related to comb filtering too. In the posts by @Wesayso, we can read that Floyd Toole assumed the comb filtering would be filled in by early reflections. But this is again a matter of dispersion pattern. So apart from its use in predicting tonality and imaging, high and low DI will at least point into the direction what is happening in with the stereo center. There is more experienced members in this domain, but I am sure the appearance of the image, its size, width, and depth position can be explained more thoroughly with objective indicators.

 

[dshreter]

As I said already, different crossover filter functions are held responsible for shifiting the position of the phantom image back and forth.
The nature of the center image is related to comb filtering too. In the posts by @Wesayso, we can read that Floyd Toole assumed the comb filtering would be filled in by early reflections. But this is again a matter of dispersion pattern. So apart from its use in predicting tonality and imaging, high and low DI will at least point into the direction what is happening in with the stereo center. There is more experienced members in this domain, but I am sure the appearance of the image, its size, width, and depth position can be explained more thoroughly with objective indicators.

That would be very interesting to have that sort of analysis. I've previously started topics here inquiring how you could measure sound stage or image focus in room, and the consensus at that time was there isn't currently a well-defined method. My assumption is that if it is hard to measure this attribute in-room, it is even harder to definite this as an attribute of a speaker.

I would be personally very interested primarily to help with measuring and optimizing system performance even more than to help guiding purchases. But either way I would like to know how to measure those qualities.

 

[Rick Sykora]

That would be very interesting to have that sort of analysis. I've previously started topics here inquiring how you could measure sound stage or image focus in room, and the consensus at that time was there isn't currently a well-defined method. My assumption is that if it is hard to measure this attribute in-room, it is even harder to definite this as an attribute of a speaker.

I would be personally very interested primarily to help with measuring and optimizing system performance even more than to help guiding purchases. But either way I would like to know how to measure those qualities.

I applaud your willingness to devote time to make progress on the SOTA speaker measurements but am not sure ASR is the place. I will discuss with Amir though. It would need some leadership and do not feel qualified.

I foresee, we will need some buckets for the improvement ideas maybe based on some set of criteria like:

1. Type of improvement (measurement, presentation, education)
1.1 Level of effort to improve?
2. How desirable the improvement is?
3. Who benefits most?

Will open up to the group to add/modify these...

 

[dwkdnvr]

As I said already, different crossover filter functions are held responsible for shifiting the position of the phantom image back and forth.
The nature of the center image is related to comb filtering too. In the posts by @Wesayso, we can read that Floyd Toole assumed the comb filtering would be filled in by early reflections. But this is again a matter of dispersion pattern. So apart from its use in predicting tonality and imaging, high and low DI will at least point into the direction what is happening in with the stereo center. There is more experienced members in this domain, but I am sure the appearance of the image, its size, width, and depth position can be explained more thoroughly with objective indicators.

Sorry, but this is just not accurate. It's not as simple as crossover order moving the image front/back, and not just looking at DI and drawing conclusions about imaging.
Image perception is still not fully understood as far as I can tell, but it is dependent on at least the following characteristics
- interaural time differences
- interaural level differences
- spectral balance of reflections vs first-arrival
- time correlations of reflections vs first-arrival
- (probably) direction of reflections

The first 2 govern lateral localization (i.e 'imaging') and there really isn't anything there that crossover topology is going to influence dramatically (at least to first order of approximation. It's possible that shallower slopes and broad overlap between drivers may interfere with the ITD/ILD perception, but the counter-examples are Dunlavy and Vandersteen which use 1st order xovers, albeit time aligned/coherent). The 3 final entries are more associated with spaciousness and soundstaging and are heavily dependent on the specifics of the room and how the speaker interacts with it and it's likely impossible to align speakers along a single 'better' to 'worse' axis independent of room any better than we already do with the preference store.
I think front/back perception is the least well understood, but I'd expect 3 and 4 to be involved. If anyone has pointers on solid discussions on this I'd be interested.

I'm all for trying to add some metrics related to spatial performance/perception, but it's certainly far more complicated than just looking at xover order and DI.

 

[dshreter]

I applaud your willingness to devote time to make progress on the SOTA speaker measurements but am not sure ASR is the place. I will discuss with Amir though. It would need some leadership and do not feel qualified.

Just to make sure my post was clear as when I re-read it I could see room for confusion. I don't have the ability to do measurements for ASR. I was trying to say I would love to learn how to measure things like sound stage, and if I knew how I would use it to optimize my own system. ASR is definitely the right place for discussing what attributes of sound are measurable though!

Now whether it is known how to measure this and if Amir would do so for reviews is a different story.