On Peer Reviewed Science

[andreasmaaan]

The DSP used was the Genelec GLM. It’s fully transparent as you can see the EQ adjustments in the software.

Ok, so EQ was the only process the DSP was doing in that case? I.e. there was no impulse response convolution or anything else more sophisticated?

 

[svart-hvitt]

Ok, so EQ was the only process the DSP was doing in that case? I.e. there was no impulse response convolution or anything else more sophisticated?

As far as I understand, the EQ is minimum phase, frequency response related (only cutting excess energy, no or extremely little adding of energy to the frequency curve). I guess the bass related EQ is where it’s most easily heard, but I would love to hear the same experiment only below/above Schroeder.

 

[Cosmik]

@andreasmaaan
But to me, this little experiment reminded me of Ockhams’s razor: It puts the theorizers’ skepticism against room DSP in another light.

I think your logic is awry:
- You and real performer/person/friend, or loudspeaker playing a recording, in any room sounds like the peformer/person/friend/recording - you always recognise their voice/sound and can tell where it's coming from. It sounds 'right'.
- Record with a microphone at your location and play it back later (headphones or speaker), and it sounds different - echoey, reverberant, hollow. It sounds 'wrong'.

Logical conclusion: being 'live' in a room with a performer/person/friend or loudspeaker playing a recording sounds 'right'. Making a recording (or measurement) with a microphone does not represent what you hear in the room.

Illogical conclusion: We need to modify the performer/person/friend or loudspeaker playing a recording while with them in a room - even though they sound 'right' to us - in order to make them/it sound right in a recording that we will never hear. To do this we will base the 'correction' i.e. (mutilation of the sound) on measurements made with a microphone.

'Correcting' what sounds 'right' while with 'the source' in a room, in order to make it sound 'right' in a recording that you know is 'wrong', means that the source will sound 'wrong' when you are with it in the room. Its wrongness will be equal to, and the inverse of, that which you heard earlier in the recording.

Think about it....

 

[andreasmaaan]

@Cosmik I think you're glossing over the important distinction between performer/person/friend on one hand, and loudspeaker on the other. The former are original sound sources, whereas the loudspeaker is a sound reproducer.

"Right" in the case of the loudspeaker therefore is anchored to a reference point that is independent of the room: the recording.

Of course, I'm also glossing over the complicating fact that the recording can only ever be reproduced in a real acoustic space (or on headphones). But believe nevertheless that the assumptions you're applying to the original sound sources (which I agree are correct in those cases) can't be simplistically assumed to extend to the reproducer (the speaker).

 

[Cosmik]

@Cosmik I think you're glossing over the important distinction between performer/person/friend on one hand, and loudspeaker on the other. The former are original sound sources, whereas the loudspeaker is a sound reproducer.

"Right" in the case of the loudspeaker therefore is anchored to a reference point that is independent of the room: the recording.

Of course, I'm also glossing over the complicating fact that the recording can only ever be reproduced in a real acoustic space (or on headphones). But believe nevertheless that the assumptions you're applying to the original sound sources (which I agree are correct in those cases) can't be simplistically assumed to extend to the reproducer (the speaker).

The speaker could be playing a dry recording of a voice, and its size and dispersion characteristics may be very similar to a human performer's head. Logically, this, in itself, would invalidate the distinction you make between a speaker playing a recording and a live performer. And we know that recordings are generally made a lot more 'dry' than sitting in the audience because of the 'hollowness ' effect - which we have discussed and probably explained in other threads.

 

[svart-hvitt]

I think your logic is awry:
- You and real performer/person/friend, or loudspeaker playing a recording, in any room sounds like the peformer/person/friend/recording - you always recognise their voice/sound and can tell where it's coming from. It sounds 'right'.
- Record with a microphone at your location and play it back later (headphones or speaker), and it sounds different - echoey, reverberant, hollow. It sounds 'wrong'.

Logical conclusion: being 'live' in a room with a performer/person/friend or loudspeaker playing a recording sounds 'right'. Making a recording (or measurement) with a microphone does not represent what you hear in the room.

Illogical conclusion: We need to modify the performer/person/friend or loudspeaker playing a recording while with them in a room - even though they sound 'right' to us - in order to make them/it sound right in a recording that we will never hear. To do this we will base the 'correction' i.e. (mutilation of the sound) on measurements made with a microphone.

'Correcting' what sounds 'right' while with 'the source' in a room, in order to make it sound 'right' in a recording that you know is 'wrong', means that the source will sound 'wrong' when you are with it in the room. Its wrongness will be equal to, and the inverse of, that which you heard earlier in the recording.

Think about it....

Real performer and recorded material playback are not the same thing. Which means you can’t easily compare the two.

Speakers are a flawed concept if one thinks it’s supposed to reproduce «real performance». Look at my avatar

I like John Watkinson (who we discussed on previous occasions) and think he has some interesting ideas. But I am not certain he has found the holy grail to bridge the gap between «real» and reproduction.

 

[oivavoi]

Some things don't change! Back to ASR after some months away, and there's still debate about room DSP! But not unreasonable, as it's probably one of the remaining areas in acoustics and psychoacoustics where the jury's still out.

Just some anecdotal 5 cents, since @svart-hvitt mentions the D&D 8C, and the room. As you know I got them a couple of months ago, and I became very enamored. Because of this "moving into the new house" project which seems to extend to eternity, I've had them in a small and suboptimal room. I compared how it was to listen to them in this room and in a larger and better room downstairs, and the difference was large. Much better in the larger room. Still, I perceived the fundamental presentation to be similar. And all my listening had to take place in the small room in the attic.

But after some time, I began to miss my previous speakers, the AVI DM10 monitors, in spite of how impressed I was by the 8Cs. The AVI's are speakers which I really really love, but it's not clear to me why I perceive them as being so good. Speaker design wise, they are not spectacular on paper. They are active and have very steep crossovers (48 db), so that's good, and the drivers are very high-performing and expensive, relative to their price point. But they are two-ways (bad because of intermodulation distortion), rectangular and boxy (bad because of diffraction), non-DSP based (bad because there's no advanced EQ), and their dipersion is not even with frequency (tweeters and woofers have different dispersion at the crossover point). Without the addition of a subwoofer, they don't have enough extension and authorithy in the bass. All of this deviates from what I believe to be best practice in speaker design. At the same time, I do perceive them to have a "clarity" and a level of detail, combined with a certain "smoothness" in their presentation, which is almost unsurpassed to my subjective ears.

Would I perceive it like this in a blind-test? I have no idea. Is it due to simple things like frequency response and dispersion pattern? Perhaps. Is it just that I have gotten used to how music sounds on them? Maybe. But anyway, this is my subjective impression. So one week ago, I finally moved the 8Cs to their new home, where they will soon be installed. In the meantime, I unpacked my old AVI's from the basement, and put them in the same listening room. I was curious as to how they would sound there compared with the 8C's. And voila, everything I had found myself secretly longing for was once again there! The unmatched clarity, the silky smoothness, the very inviting way of presenting music. The basic character of the AVI's, which I had heard in the previous setup with them and really enjoyed, was there in this new room as well. I have found myself enjoying music even more with the AVI's than with the 8Cs in this room, even though the 8Cs easily surpass them in some areas - better imaging/soundstaging farther away from the speakers, more bass extension and superior dynamics.

Why? Perhaps it's due to the bass extension of the 8C's, which may overpower such a small room. Or perhaps it's that the AVI's are bass light, which may make the higher frequencies appear "clearer". Or who knows, perhaps the engineer at AVI really know what he's doing. According to AVI, using a waveguide will by necessity color the sound, and they think that anything that messes ever so slightly with the direct sound from the drivers is a trade-off that is not worth it, and the overriding concern in speaker design should be to optimize how it sounds on-axis. Personally I'm also not sure about using Class D amps for the high frequencies. Whatever the reason, I do perceive a difference between the 8Cs and the AVI's, that is there independent of the room.

This is all anecdotal of course. But my personal experience so far indicates that we can indeed separate out the room when we listen, and that speakers will have their "sound" even in different rooms. Just like instruments btw. This isn't to say that room eq will necessarily be detrimental. I have heard good results with room eq/DSP, and I have heard corrections I didn't like. I think a large reason for why room DSP may sound subjectively good is that it performs "speaker EQ", which can be important for bad speakers, and that it can take care of level and distance differences between speakers. Beyond that, whether it's good or not good to adjust the direct sound in order to adjust for irregularities in the indirect sound, on that I'm still agnostic.

 

[svart-hvitt]

Some things don't change! Back to ASR after some months away, and there's still debate about room DSP! But not unreasonable, as it's probably one of the remaining areas in acoustics and psychoacoustics where the jury's still out.

Just some anecdotal 5 cents, since @svart-hvitt mentions the D&D 8C, and the room. As you know I got them a couple of months ago, and I became very enamored. Because of this "moving into the new house" project which seems to extend to eternity, I've had them in a small and suboptimal room. I compared how it was to listen to them in this room and in a larger and better room downstairs, and the difference was large. Much better in the larger room. Still, I perceived the fundamental presentation to be similar. And all my listening had to take place in the small room in the attic.

But after some time, I began to miss my previous speakers, the AVI DM10 monitors, in spite of how impressed I was by the 8Cs. The AVI's are speakers which I really really love, but it's not clear to me why I perceive them as being so good. Speaker design wise, they are not spectacular on paper. They are active and have very steep crossovers (48 db), so that's good, and the drivers are very high-performing and expensive, relative to their price point. But they are two-ways (bad because of intermodulation distortion), rectangular and boxy (bad because of diffraction), non-DSP based (bad because there's no advanced EQ), and their dipersion is not even with frequency (tweeters and woofers have different dispersion at the crossover point). Without the addition of a subwoofer, they don't have enough extension and authorithy in the bass. All of this deviates from what I believe to be best practice in speaker design. At the same time, I do perceive them to have a "clarity" and a level of detail, combined with a certain "smoothness" in their presentation, which is almost unsurpassed to my subjective ears.

Would I perceive it like this in a blind-test? I have no idea. Is it due to simple things like frequency response and dispersion pattern? Perhaps. Is it just that I have gotten used to how music sounds on them? Maybe. But anyway, this is my subjective impression. So one week ago, I finally moved the 8Cs to their new home, where they will soon be installed. In the meantime, I unpacked my old AVI's from the basement, and put them in the same listening room. I was curious as to how they would sound there compared with the 8C's. And voila, everything I had found myself secretly longing for was once again there! The unmatched clarity, the silky smoothness, the very inviting way of presenting music. The basic character of the AVI's, which I had heard in the previous setup with them and really enjoyed, was there in this new room as well. I have found myself enjoying music even more with the AVI's than with the 8Cs in this room, even though the 8Cs easily surpass them in some areas - better imaging/soundstaging farther away from the speakers, more bass extension and superior dynamics.

Why? Perhaps it's due to the bass extension of the 8C's, which may overpower such a small room. Or perhaps it's that the AVI's are bass light, which may make the higher frequencies appear "clearer". Or who knows, perhaps the engineer at AVI really know what he's doing. According to AVI, using a waveguide will by necessity color the sound, and they think that anything that messes ever so slightly with the direct sound from the drivers is a trade-off that is not worth it, and the overriding concern in speaker design should be to optimize how it sounds on-axis. Personally I'm also not sure about using Class D amps for the high frequencies. Whatever the reason, I do perceive a difference between the 8Cs and the AVI's, that is there independent of the room.

This is all anecdotal of course. But my personal experience so far indicates that we can indeed separate out the room when we listen, and that speakers will have their "sound" even in different rooms. Just like instruments btw. This isn't to say that room eq will necessarily be detrimental. I have heard good results with room eq/DSP, and I have heard corrections I didn't like. I think a large reason for why room DSP may sound subjectively good is that it performs "speaker EQ", which can be important for bad speakers, and that it can take care of level and distance differences between speakers. Beyond that, whether it's good or not good to adjust the direct sound in order to adjust for irregularities in the indirect sound, on that I'm still agnostic.

Did you try maximizing the direct sound when listening to A vs B? Maximizing DS means minimizing room contribution. So it may be worth a test? How near can you sit A and B?

 

[andreasmaaan]

Interesting post @oivavoi! You don't happen to have measurements of the DM10 do you?

I'd be curious to see how they measure to try to nut out the causes of the perceived differences in sound between them and the 8C.

And would you be able to give a brief description of the room?

 

[oivavoi]

@svart-hvitt : this was not an A/B test! Only based on my fallible long term acoustic memory. So very unscientific and not authoritative in any way. My experience was that I need to sit closer, with the speakers closer to each other, with the AVIs than with the 8Cs. I assume that's due to the more controlled and somewhat narrower dispersion of the 8Cs, which makes them image better at a distance. So in both cases I tried to find a listening and speaker positin which gave me a ratio of direct to indirect sound which made the imaging clear.

@andreasmaaan: unfortunately, no measurements at the moment! I can see if I can do simple measurements tomorrow. AVI haven't published detailed frequency plots or polars, which is another thing I don't regard as best practice speaker-wise. My hunch would be the following: I would guess that they might be ever slightly hot in the midrange and/or the treble, which could give a sense of greater clarity. That's the non-charitable interpretation of how I perceive their performance. I would also guess that they would perform very well on advanced distorsion tests (i.e. not static signals, but dynamic material with transients etc). My impression is that they are very "clean", there is no noise etc which masks low-level detail. So the signal stops when it should stop, and starts when it should start. The room: approx 5 x 3.5 m, with a very low sloping ceiling. So far from optimal acoustically.

Another thought I have had is this, which is related to the topic of the thread: Do we really know all that is to know about speaker performance, based on the usual suspects of (peer reviewed) parameters? I just read the long thread on avsforum on the blind shootout between M2 and the Salon 2. The M2 measures better. Hands-down. It's active to boot. Still, the Salon 2 beat it rather handily in the comparison. Many people desribed the Salon 2 as more "smooth" and "silky" - some of the some attributes that I appreciate in the DM10s. Many people on Avsforum attributed this to the difference in dispersion, the Salon 2 have somewwhat wider dispersion. But is that all there is to it? I'm not sure. (apologies if this qualifies as creating subjectivist FUD on speaker design)

 

[andreasmaaan]

Another thought I have had is this, which is related to the topic of the thread: Do we really know all that is to know about speaker performance, based on the usual suspects of (peer reviewed) parameters?

Ha, I think it depends on how you define "all there is to know". We can certainly take extremely detailed measurements of speakers in much the same way we can any electronic component.

What we have less data on (probably simply because there are more variables) are the correlations between measurements and subjective experience. Some parameters (e.g. dispersion) are more difficult to study, because unlike electronics, they can't be artificially applied to a signal electronically for optimal comparative listening studies. So often the listening studies necessarily have to test multiple variables simultaneously.

However, many aspects of how speaker measurements correlate with subjective experience have been studied in some detail and are reasonably well understood.

In terms of your preference for the DM10 over the 8C, we can only speculate, especially as we have only limited measurements of the 8C and no measurements of the DM10. But my suspicion is that much of the subjective difference would come down to the 8C's controlled directivity creating a significantly different reverberant sound field in-room, i.e. a reverberant field with far less bass and midrange energy than a non-cardioid speaker; this may not be subjectively as pleasing. This is just my hunch though..

 

[andreasmaaan]

The M2 measures better. Hands-down.

I've seen basic measurements of both speakers but have not looked into it in depth. In what respects does the M2 measure significantly better?

 

[svart-hvitt]

@svart-hvitt : this was not an A/B test! Only based on my fallible long term acoustic memory. So very unscientific and not authoritative in any way. My experience was that I need to sit closer, with the speakers closer to each other, with the AVIs than with the 8Cs. I assume that's due to the more controlled and somewhat narrower dispersion of the 8Cs, which makes them image better at a distance. So in both cases I tried to find a listening and speaker positin which gave me a ratio of direct to indirect sound which made the imaging clear.

@andreasmaaan: unfortunately, no measurements at the moment! I can see if I can do simple measurements tomorrow. AVI haven't published detailed frequency plots or polars, which is another thing I don't regard as best practice speaker-wise. My hunch would be the following: I would guess that they might be ever slightly hot in the midrange and/or the treble, which could give a sense of greater clarity. That's the non-charitable interpretation of how I perceive their performance. I would also guess that they would perform very well on advanced distorsion tests (i.e. not static signals, but dynamic material with transients etc). My impression is that they are very "clean", there is no noise etc which masks low-level detail. So the signal stops when it should stop, and starts when it should start. The room: approx 5 x 3.5 m, with a very low sloping ceiling. So far from optimal acoustically.

Another thought I have had is this, which is related to the topic of the thread: Do we really know all that is to know about speaker performance, based on the usual suspects of (peer reviewed) parameters? I just read the long thread on avsforum on the blind shootout between M2 and the Salon 2. The M2 measures better. Hands-down. It's active to boot. Still, the Salon 2 beat it rather handily in the comparison. Many people desribed the Salon 2 as more "smooth" and "silky" - some of the some attributes that I appreciate in the DM10s. Many people on Avsforum attributed this to the difference in dispersion, the Salon 2 have somewwhat wider dispersion. But is that all there is to it? I'm not sure. (apologies if this qualifies as creating subjectivist FUD on speaker design)

Sure, no AB, and totally unscientific - which is, however, how science always starts (that’s for another discussion though...).

It seems to me, based on what I experienced from the GLM room DSP experiment, that the differences you talk about could come down to in-room imperfections. Since you know some knowledgeable DSP guys, you should try and see what they can do with your speaker response. It’s foolish to write off what DSP can do (I don’t think you do that), and I am quite sure it’s beneficial in LFs (where I think «muddiness» and «lack of detail» may origin in some cases).

 

[oivavoi]

I've seen basic measurements of both speakers but have not looked into it in depth. In what respects does the M2 measure significantly better?

Flatter frequency response and more smooth and controlled off-axis. But this is also from memory - I may remember wrongly!

 

[svart-hvitt]

Ha, I think it depends on how you define "all there is to know". We can certainly take extremely detailed measurements of speakers in much the same way we can any electronic component.

What we have less data on (probably simply because there are more variables) are the correlations between measurements and subjective experience. Some parameters (e.g. dispersion) are more difficult to study, because unlike electronics, they can't be artificially applied to a signal electronically for optimal comparative listening studies. So often the listening studies necessarily have to test multiple variables simultaneously.

However, many aspects of how speaker measurements correlate with subjective experience have been studied in some detail and are reasonably well understood.

In terms of your preference for the DM10 over the 8C, we can only speculate, especially as we have only limited measurements of the 8C and no measurements of the DM10. But my suspicion is that much of the subjective difference would come down to the 8C's controlled directivity creating a significantly different reverberant sound field in-room, i.e. a reverberant field with far less bass and midrange energy than a non-cardioid speaker; this may not be subjectively as pleasing. This is just my hunch though..

On the issue of masurements and our understanding of audio:

If you have a sailing ship, and all measurements are built around the habit of having to do with sailing ships, what would those measurements tell you about the first motorized ships? When measurements are built around old habits, legacy design, a naive reader of measurements could conclude that the sailing ships are better; they measure better in tests that are based on our habit of seeing sailing ships.

A better mind - as opposed to a lab slave - would see that a more innovative design may be a step forward despite poor measurements of the new design. This raises the question what is the most important: The better design or the better measurements? I would choose a poorly measuring motor car every day over a brilliantly measuring horse-driven car.

In mature technologies, measurements may tell us all there is to know. Speakers are not mature in that sense, I believe. So speakers may be one area where the more intelligent design combined with good implementation (quality control) are more important than measurements. This is good news for research-driven companies and for those that appreciate good design and implementation.

 

[andreasmaaan]

Flatter frequency response and more smooth and controlled off-axis. But this is also from memory - I may remember wrongly!

Ok interesting, thanks.

Have had a quick look at the measurements that are available online and actually both speakers seem to have about equally flat on-axis response and very even and well-controlled off-axis response (although I can't find independent polar response measurements for the M2 so I'm trusting JBL on this one).

They are very different speakers though - the Revels are generally wider dispersion as you mentioned.

I'd expect the M2's to be lower distortion given the drivers used, but it's hard to say..

These are the measurements I found in case you're interested:

• http://www.audioheritage.org/vbulletin/showthread.php?33972-JBL-Master-Reference-Monitor/page73
• https://www.soundstagenetwork.com/measurements/speakers/revel_ultima_salon2/

 

[andreasmaaan]

In mature technologies, measurements may tell us all there is to know. Speakers are not mature in that sense, I believe. So speakers may be one area where the more intelligent design combined with good implementation (quality control) are more important than measurements. This is good news for research-driven companies and for those that appreciate good design and implementation.

In what sense are speakers not a mature technology in your opinion?

What new designs that do not measure well are promising in your opinion? In what ways do the measurements not tell the full story?

 

[svart-hvitt]

In what sense are speakers not a mature technology in your opinion?

What new designs that do not measure well are promising in your opinion? In what ways do the measurements not tell the full story?

Astonishingly, very few speaker designers have jumped into the rabbit hole* of point source. I think this is due to the following:

1) Lack of vision, lack of wish to explore, lack of ability to be idea oriented.

2) Lack of skill, competency.

3) Lack of time.

4) Lack of financial resources.

John Watkinson has 1, 2 and 3. I haven’t heard his design but am certain he is short of 4.

Genelec decided to scrap the point source idea many years ago. But then they decided to brush dust off the (very!) old idea again some years ago. In the meantime, Genelec had perfected enclosure production (die cast aluminium) and were able to control small, yet important details on an assembly line.

On some parameters, the cheaper Genelecs measure better than the more complex and costlier coaxials. So people wonder if the cheaper, older ones are better than the more expensive ones. This may be a good example to illustrate the point: Should you go for (certain) measurements only or look at the overall design idea (in combination with listening)?


*I like to use «rabbit hole» in a positive light. To me rabbit hole means being explorative, adventurous, risk-seeking in a good way. For some reason or another, I have the impression that native English speaking people think of rabbit hole as something negative; waste of time and resources. That’s a sad interpretation of rabbit hole, I think.

 

[andreasmaaan]

@svart-hvitt could you please be specific. What's one older Genelec model that measures better than one newer Genelec model that you suspect might sound better?

Hard to have this kind of discussion in the abstract.

For point source speakers, have you checked out the Danley Synergy range? Those are very interesting from an engineering perspective.

 

[svart-hvitt]

@svart-hvitt could you please be specific. What's one older Genelec model that measures better than one newer Genelec model that you suspect might sound better?

Hard to have this kind of discussion in the abstract.

For point source speakers, have you checked out the Danley Synergy range? Those are very interesting from an engineering perspective.

This is what I meant about old measuring better than new:

https://www.community.genelec.com/forum/-/message_boards/view_message/1004384#_19_message_1004384

I will look up Danley.

EDIT: I looked up Danley on a previous occasion but didn’t understand the concept then. Will try again.