- Joined
- Nov 6, 2018
- Messages
- 1,469
- Likes
- 4,895
Don't care if model is valid as long as we get pretty pictures!
Wonderful dynamic visualization. Thank you.
-
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!
Complaint thread about speaker measurements
- Thread starter amirm
- Start date
Don't care if model is valid as long as we get pretty pictures!
Wonderful dynamic visualization. Thank you.
tuga
Major Contributor
Heart rate looks stable, I think he's gonna make it.
Regarding the relevance of controlled directivity for near field listening, the same movement close to the speaker changes the angle of the path of the direct sound relative the acoustic axis much more than when far away. While reducing the listening distance shrinks the space where tonality - and consequently imaging - are consistent, good horizontal directivity not only increases the width but also the depth of it, while good vertical directivity increases its height.
Further, in the near field, the strongest reflections may well be from the floor and ceiling or a desk or some other flat horizontal surface. For a 1 m listening distance, for 0.75-1.5 m to floor or ceiling, the reflected sound is emitted around 55-70 degrees above or below the acoustic axis, while being attenuated 5-10 dB and being delayed between 2-6 ms. For a desk where the surface is 0.5 m below the acoustic axis, for 0.5-1.5 m listening distance, the angle is around 35-75 degrees, while the attenuation is only 2-7 dB and the delay between 0.6-4 ms. The spectrum of those reflections likely matters a fair bit.
Professional audio companies such as Genelec and Neumann mention a consistent and wide - and therefore also deep - listening area and controlled reflections from near vertical objects (such as consoles) and the ceiling as one of the primary benefits of the controlled directivity of their speakers.
Thus, speakers that do well on the spinorama will also likely sound good at closer listening distances - if not better.
Further, in the near field, the strongest reflections may well be from the floor and ceiling or a desk or some other flat horizontal surface. For a 1 m listening distance, for 0.75-1.5 m to floor or ceiling, the reflected sound is emitted around 55-70 degrees above or below the acoustic axis, while being attenuated 5-10 dB and being delayed between 2-6 ms. For a desk where the surface is 0.5 m below the acoustic axis, for 0.5-1.5 m listening distance, the angle is around 35-75 degrees, while the attenuation is only 2-7 dB and the delay between 0.6-4 ms. The spectrum of those reflections likely matters a fair bit.
Professional audio companies such as Genelec and Neumann mention a consistent and wide - and therefore also deep - listening area and controlled reflections from near vertical objects (such as consoles) and the ceiling as one of the primary benefits of the controlled directivity of their speakers.
Thus, speakers that do well on the spinorama will also likely sound good at closer listening distances - if not better.
D
Deleted member 16543
Guest
I've missed many replies while hastily posting, and then I got banned (WTF?) and couldn't reply individually.
I think starting from scratch might be better than going back and address little things here and there.
I can't speak for other people's grievances about the Klippel's system (nor do I care), so these are just my personal observations about something that is -can I say it?- 'missing' in those measurements.
Missing might not even be the right word, as those measurements are not missing anything, as far as the speaker in and of itself goes. And since they don't even really intend to go much beyond a clinical description of just the speaker being measured, if not for the PIR curve, asking them to do just so might even be a little too demanding on my part.. So, technically, the spinorama is not missing anything of what's it's trying to portray.
I write this as an attempt to settle the matter from that point of view. I have nothing against spinoramas, to the extent of what they try to do.
At the same time I do believe that what they portray is not exactly the full picture, from the point of view of human's perception of sound balance. There is data and then there is interpretation of the data.. and then there is correlation from this data to how humans hear.
So, if you read nothing else but these first few words, the following is the core of my observations about the spinorama measurements, for near field listening:
At the end of the day, I'm just trying to find something (i.e. a target curve) that has as much a strong correlation to the perceived sound of a near field speaker at the sweet spot, as the PIR has for a listener in the room's far field.
Here comes the boring part, if you care to read further.
Some definitions first.
1. Near field. In the sense I have been using this term all along, this indicates the region of space where the power from the first wave front coming from the speakers is higher than the power of the reflected sound caused by room reflections. At least in a time windowing framework that correlates to how we perceive sounds (psychoacoustic gating).
It is NOT the definition related to the speaker dimensions and wavelength. In fact, I assume that we are always in the far field from that point of view (above a certain frequency, of course), since being closer than that will likely produce too high variations in frequency response with minimal head movements.
2. Measurements. There are many types. I personally use one with variable, decreasing with frequency windowing. The reasoning being accounting for about the same number of cycles (direct+reflected) when measuring in room. This is the psychoacoustics algorithm used in Acourate, if anybody wants to dig deeper.
Nobody has ever talked about steady state response (at least not me). Putting words into people's mouth just to make up arguments to attack is childish behavior, and I was hoping that in a site called Audio SCIENCE Review I would manage to avoid the pack of bootlickers types. Alas...
On to the argumentation.
Imagine an anechoic situation, where all we get is the direct sound from the speakers. It doesn't matter how we gate the measurements, because the sound passes through once and then it's gone forever behind us without reflections (I am also aware that this is a theoretical situation, since no regular room is anechoic.. under a certain frequency not even anechoic rooms!).
Now we keep the speakers/measurement mic mutual geometry the same and add a room around. We just created reflections as a direct consequence, and gating the measurements is very important now. Same speakers, same room, same measuring point can get us quite different results depending just on gating/windowing.
Since we are making measurements and trying to correlate them to how humans perceive the sound balance (isn't the end goal that of telling how balanced a speaker will sound?), it makes most sense to me to measure according to some psychoacoustic principle, hence the Acourate software I use.
I don't think I'm saying anything heretical when I say that if we were to put our head in place of the measurement microphone in these two types of scenarios, we would perceive a very different sound balance. This is coarsely related to the topic of mic free field vs. reverberant field equalization, for example. Nothing new so far.
What tools do we have to correlate this different sound perception to some kind of measurement/target curve (other than using a psychoacoustic algorithm and in situ measurement)? The spinorama's on-axis, listening window, room reflection, power and DI curves would be the same in both scenarios (since they don't depend on the listening environment), so none of those curves is the tool we're looking for.
In the scenario where the room is present, the answer to this question seems to be (give or take, of course) the predicted in room response. But when we move closer, into the near field, the answer is not as easily found in nearing a curve provided by the spinorama to a certain target, whatever that target might be and whatever gating we might decide to use.
Now, some people might argue that PIR is only one of the many curves, and that all other curves need to look right for the speakers to sound balanced. But do they really? Can this be assumed as a rule that is set in stone? Even for instances where, for example, there's not even a proper 'listening window' (not as wide as the one averaged by the spinorama, at least) but the intended use of a particular set of speakers (like near field studio monitors) is that of sounding 'correct' just in a specific point in space, i.e. the sweet spot?
People that have read this are probably familiar with the concept that (using a certain gating that correlates to how human perceive sound) whatever actual measurement we start with, we can always EQ the speakers so that, at the intended listening position, the psychoacoustically processed measured response can be matched to a certain target response. The acoustic end results will be very similar (within reasonable expectations, of course) starting from very different types of speakers, that measure quite differently initially, once the psychoacoustic responses are made to match the same target response.
So, if one can EQ different speakers to make them sound more or less equivalent (at the listening position and only there!), isn't that the equivalent of modifying the on axis response (and all the other directions' responses as well) by the same EQ weighting factor?
If we were to measure the DSP+speaker system with Klippel's apparatus, the software -not knowing any better- would measure an on-axis response very different (and not flat at all) than just the speaker's own response alone.
Yet, combined with the speaker's directivity characteristics and the room they are in, that non flat on axis response is part of a DSP+speakers system that, at the end of the day, sounds 'right' at the listening position we measured and EQed for.
So, this to me is proof that a flat on axis response is not always necessary to achieve a good balance in sound.
All this seems like pretty simple logical reasoning to me (confirmed by experience, incidentally), and probably a key stone concept that, if not at least pondered (and ideally agreed upon), will prevent from ever getting to the bottom of the matter.. So, since this is probably starting to get to that tl;dr threshold for many, I'll stop here for now and make sure whoever is really interested in hearing what I have to say has time to ask questions and clarify where I have been less than clear.
I think starting from scratch might be better than going back and address little things here and there.
I can't speak for other people's grievances about the Klippel's system (nor do I care), so these are just my personal observations about something that is -can I say it?- 'missing' in those measurements.
Missing might not even be the right word, as those measurements are not missing anything, as far as the speaker in and of itself goes. And since they don't even really intend to go much beyond a clinical description of just the speaker being measured, if not for the PIR curve, asking them to do just so might even be a little too demanding on my part.. So, technically, the spinorama is not missing anything of what's it's trying to portray.
I write this as an attempt to settle the matter from that point of view. I have nothing against spinoramas, to the extent of what they try to do.
At the same time I do believe that what they portray is not exactly the full picture, from the point of view of human's perception of sound balance. There is data and then there is interpretation of the data.. and then there is correlation from this data to how humans hear.
So, if you read nothing else but these first few words, the following is the core of my observations about the spinorama measurements, for near field listening:
At the end of the day, I'm just trying to find something (i.e. a target curve) that has as much a strong correlation to the perceived sound of a near field speaker at the sweet spot, as the PIR has for a listener in the room's far field.
Here comes the boring part, if you care to read further.
Some definitions first.
1. Near field. In the sense I have been using this term all along, this indicates the region of space where the power from the first wave front coming from the speakers is higher than the power of the reflected sound caused by room reflections. At least in a time windowing framework that correlates to how we perceive sounds (psychoacoustic gating).
It is NOT the definition related to the speaker dimensions and wavelength. In fact, I assume that we are always in the far field from that point of view (above a certain frequency, of course), since being closer than that will likely produce too high variations in frequency response with minimal head movements.
2. Measurements. There are many types. I personally use one with variable, decreasing with frequency windowing. The reasoning being accounting for about the same number of cycles (direct+reflected) when measuring in room. This is the psychoacoustics algorithm used in Acourate, if anybody wants to dig deeper.
Nobody has ever talked about steady state response (at least not me). Putting words into people's mouth just to make up arguments to attack is childish behavior, and I was hoping that in a site called Audio SCIENCE Review I would manage to avoid the pack of bootlickers types. Alas...
On to the argumentation.
Imagine an anechoic situation, where all we get is the direct sound from the speakers. It doesn't matter how we gate the measurements, because the sound passes through once and then it's gone forever behind us without reflections (I am also aware that this is a theoretical situation, since no regular room is anechoic.. under a certain frequency not even anechoic rooms!).
Now we keep the speakers/measurement mic mutual geometry the same and add a room around. We just created reflections as a direct consequence, and gating the measurements is very important now. Same speakers, same room, same measuring point can get us quite different results depending just on gating/windowing.
Since we are making measurements and trying to correlate them to how humans perceive the sound balance (isn't the end goal that of telling how balanced a speaker will sound?), it makes most sense to me to measure according to some psychoacoustic principle, hence the Acourate software I use.
I don't think I'm saying anything heretical when I say that if we were to put our head in place of the measurement microphone in these two types of scenarios, we would perceive a very different sound balance. This is coarsely related to the topic of mic free field vs. reverberant field equalization, for example. Nothing new so far.
What tools do we have to correlate this different sound perception to some kind of measurement/target curve (other than using a psychoacoustic algorithm and in situ measurement)? The spinorama's on-axis, listening window, room reflection, power and DI curves would be the same in both scenarios (since they don't depend on the listening environment), so none of those curves is the tool we're looking for.
In the scenario where the room is present, the answer to this question seems to be (give or take, of course) the predicted in room response. But when we move closer, into the near field, the answer is not as easily found in nearing a curve provided by the spinorama to a certain target, whatever that target might be and whatever gating we might decide to use.
Now, some people might argue that PIR is only one of the many curves, and that all other curves need to look right for the speakers to sound balanced. But do they really? Can this be assumed as a rule that is set in stone? Even for instances where, for example, there's not even a proper 'listening window' (not as wide as the one averaged by the spinorama, at least) but the intended use of a particular set of speakers (like near field studio monitors) is that of sounding 'correct' just in a specific point in space, i.e. the sweet spot?
People that have read this are probably familiar with the concept that (using a certain gating that correlates to how human perceive sound) whatever actual measurement we start with, we can always EQ the speakers so that, at the intended listening position, the psychoacoustically processed measured response can be matched to a certain target response. The acoustic end results will be very similar (within reasonable expectations, of course) starting from very different types of speakers, that measure quite differently initially, once the psychoacoustic responses are made to match the same target response.
So, if one can EQ different speakers to make them sound more or less equivalent (at the listening position and only there!), isn't that the equivalent of modifying the on axis response (and all the other directions' responses as well) by the same EQ weighting factor?
If we were to measure the DSP+speaker system with Klippel's apparatus, the software -not knowing any better- would measure an on-axis response very different (and not flat at all) than just the speaker's own response alone.
Yet, combined with the speaker's directivity characteristics and the room they are in, that non flat on axis response is part of a DSP+speakers system that, at the end of the day, sounds 'right' at the listening position we measured and EQed for.
So, this to me is proof that a flat on axis response is not always necessary to achieve a good balance in sound.
All this seems like pretty simple logical reasoning to me (confirmed by experience, incidentally), and probably a key stone concept that, if not at least pondered (and ideally agreed upon), will prevent from ever getting to the bottom of the matter.. So, since this is probably starting to get to that tl;dr threshold for many, I'll stop here for now and make sure whoever is really interested in hearing what I have to say has time to ask questions and clarify where I have been less than clear.
Last edited by a moderator:
Thank you for your question. We consider the W371 as a member of the Ones family and therefore we have not considered to support other high performance monitors, such as the S360, for the time being. We will carefully keep following up the requests we receive from our customers and if we see there is such a need in the future, we will consider the other main monitors to be supported by the W371.
@amirm, could you please start measuring hiss for all active speakers, like this maybe?
https://www.audiosciencereview.com/...-lp-6-review-studio-monitor.17978/post-654580
This is very important as a high hiss can be a deal breaker.
Thanks.
https://www.audiosciencereview.com/...-lp-6-review-studio-monitor.17978/post-654580
This is very important as a high hiss can be a deal breaker.
Thanks.
ROOSKIE
Major Contributor
We have seen examples here where the deviations in on axis response were "flattened" out/Compensated for in the PIR due the interactions with the predicted room bounce/reflections.
I think this is understood as a possibility and thus non-flat on axis anechoic sometimes works out, sometime flat actually doesn't due to the same effects .
I will say that the PIR never quite matches to a "T", what I have measured in my actual room. There is a very meaningful deviation. I always expect to measure in my room and can no longer value the opinions of those who are not doing this.
I basically trust the Kipple testing and the theories presented here enough to start with speakers that have performed well here. Doing this I have learned a lot and have excellent sound.
That said, with speakers, nothing measured here is a fact, rather "good directions for how to get where I am going". There are surely other ways of course and most importantly nothing trumps what I actually measure and hear in my room.
I can say this. I started working with DIY pro drivers and huge waveguides, to learn/explore a bit I purchased a pair of Monoprice SR10 passive PA speakers with 10"woofers and 1" compression driver - $120 A PAIR. The cheapest I could find with decent reviews.
Pre in room PEQ= NOPE, muddy and muddy and muddy and with very elevated lower mids/upper midbass and very peaky in the 200-1.5k range.
Post in room PEQ = Wow subjectively & nearly match other PEQ'd speakers even ones with very good directivity.
I matched the in room listening window response (MMM) closely to both my EQ'd JBL 530 and JBL A130's, using the help of stereo bass woofers and my general target curve. (a slight variation of the Harman downward slope)
These 3 speaker pairs all have very equal perceived sound quality at this point. All sound excellent.(All three speaker sets do have distinct qualitative differences. They do sound slightly/meaningfully different from one another - sighted at least)
With PEQ and some careful measurements and close attention to measured distortion levels and subwoofer integration this is a very accurate system. Can it be beaten? YES I am sure it can & simultaneously these two black hunks of plastic sound pretty amazing, with very low Harmonic distortion due to using some very decent(surprisingly so) PA drivers built to handle SPL.
In addition to my anecdotal notes above, speakers have been tested here that did not have flat on axis responses but seemed to respond very favorably to PEQ (sans the flat anechoic, generally with such PEQ'd speakers directivity has been noted to be good such as the Revel M55XC)
Last edited:
D
Deleted member 16543
Guest
@ROOSKIE. That's not very difficult to believe for me at all..
It still relates more to the far field listening conditions camp. But it ties with the near field listening conditions too, according to the very parts of my post you cited.
I do think that by using Acourate's algorithm the three speakers would probably get perceptually even closer to each other (PEQ can only get you that close to a target curve)..
Now let's imagine an experiment. What happens as you move closer and closer to those three speakers when you listen? Don't they start to sound more and more different? They all will sound brighter, as a general rule, but not exactly in the same way. The balance (although generally brighter for all of them) will now be much more different.
That's because by getting closer the on axis response is more prevalent than the reflected sound. And the on axis response is probably very different among the three speakers. Also, HOW that response is measured is very important.
Now imagine that one of the three speakers was a pair that's specifically designed for near field listening and sounded great off the shelf.
Assuming you could get them to reach a decent SPL, by equalizing that pair for far field listening, you have changed the on axis response so that when listened to again in near field they don't sound great anymore at all.
And that's all I'm saying: great sounding speakers in far field don't sound great at all in near field, and vice versa.
I really don't understand why this very simple and self evident fact was taken so controversially by some people.
To me, an impossibly ideal speaker would have a tilted down on axis response (to sound balanced in the near field), and off axis response slightly tilted up in respect to the off axis response of a speaker with ideal PIR (to account for the sloping down on axis response), so that it would work in far field too.
For physics reasons, getting the first right prevents the second to happen without employing extra drivers firing off axis.
And employing extra drivers that considerably overlap in the frequency domain is always a recipe for comb filtering in the near field.
So I don't see such a speaker being available for sale anytime soon.
Maybe there's a simple (although quite compromised, how else could it be?) solution, for active speakers.
A switch that applies the near field target response EQ for when you listen in near field, for a speaker that has great spinorama measurements for far field listening.
I can see the Kii 3s working somewhat fine for such a concept.
But then of course there's other stuff related to near field listening that the Kii 3s are not designed to perfectly accomplish.
A lot of aspects that are secondary in far field become of the most importance in near field, and vice versa.
Phase response, cabinet diffraction, cabinet resonance (to a degree), ported vs. sealed cabinet, point source, etc.. are all aspects that are not so important in far field as they are in near field.
And there's no other way to address those than choosing certain physical characteristics of how a speaker is built. No DSP can possibly fix them.
I'm not sure how the coaxial genelecs measure as far as PIR, but if they measure well I can see them being a perfect candidate for both great near and far field listening (by employing the near field EQ switch).
They address almost all of the points that are important for near field listening by sheer cabinet design and drivers choice, while having the likely good directivity of 3 way speakers.
It still relates more to the far field listening conditions camp. But it ties with the near field listening conditions too, according to the very parts of my post you cited.
I do think that by using Acourate's algorithm the three speakers would probably get perceptually even closer to each other (PEQ can only get you that close to a target curve)..
Now let's imagine an experiment. What happens as you move closer and closer to those three speakers when you listen? Don't they start to sound more and more different? They all will sound brighter, as a general rule, but not exactly in the same way. The balance (although generally brighter for all of them) will now be much more different.
That's because by getting closer the on axis response is more prevalent than the reflected sound. And the on axis response is probably very different among the three speakers. Also, HOW that response is measured is very important.
Now imagine that one of the three speakers was a pair that's specifically designed for near field listening and sounded great off the shelf.
Assuming you could get them to reach a decent SPL, by equalizing that pair for far field listening, you have changed the on axis response so that when listened to again in near field they don't sound great anymore at all.
And that's all I'm saying: great sounding speakers in far field don't sound great at all in near field, and vice versa.
I really don't understand why this very simple and self evident fact was taken so controversially by some people.
To me, an impossibly ideal speaker would have a tilted down on axis response (to sound balanced in the near field), and off axis response slightly tilted up in respect to the off axis response of a speaker with ideal PIR (to account for the sloping down on axis response), so that it would work in far field too.
For physics reasons, getting the first right prevents the second to happen without employing extra drivers firing off axis.
And employing extra drivers that considerably overlap in the frequency domain is always a recipe for comb filtering in the near field.
So I don't see such a speaker being available for sale anytime soon.
Maybe there's a simple (although quite compromised, how else could it be?) solution, for active speakers.
A switch that applies the near field target response EQ for when you listen in near field, for a speaker that has great spinorama measurements for far field listening.
I can see the Kii 3s working somewhat fine for such a concept.
But then of course there's other stuff related to near field listening that the Kii 3s are not designed to perfectly accomplish.
A lot of aspects that are secondary in far field become of the most importance in near field, and vice versa.
Phase response, cabinet diffraction, cabinet resonance (to a degree), ported vs. sealed cabinet, point source, etc.. are all aspects that are not so important in far field as they are in near field.
And there's no other way to address those than choosing certain physical characteristics of how a speaker is built. No DSP can possibly fix them.
I'm not sure how the coaxial genelecs measure as far as PIR, but if they measure well I can see them being a perfect candidate for both great near and far field listening (by employing the near field EQ switch).
They address almost all of the points that are important for near field listening by sheer cabinet design and drivers choice, while having the likely good directivity of 3 way speakers.
Last edited by a moderator:
Similar threads
- Poll