• 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!

Serious Question: How can DAC's have a SOUND SIGNATURE if they measure as transparent? Are that many confused?

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Also curious to hear from @TomB19, @raistlin65, and others:

What in your opinion is the ideal (in terms of listener preference) frequency response and directional characteristics for a loudspeaker?

And what evidence is your view based on?

I'd really like to know of alternatives to the Olive studies on this question. As mentioned above, I have a lot of reservations about the Olive research, but I know of little alternative data.
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
What in your opinion is the ideal (in terms of listener preference) frequency response and directional characteristics for a loudspeaker?


It's an indisputable fact that the Harman data was collected in a scientific manner (evidence-based) but I don't see how "average listener preference" relates to accurate reproduction of the recorded signal. In my view it merely provides information regarding people's tastes.


Regarding frequency response, most people don't listen to recordings of acoustic music and some have never listened to a single live unamplified concert in their life; "realism" is not a requirement for the majority. Besides it is impossible to determine "realism" unless one is familiar with live and uses long term listening assessment and adequate recordings. An short-duration AB comparison is totally unfit for such purpose.

I listen mostly to classical music plus a bit of jazz, not to Krell-Barbie-Pigeon-Bubbles nor pop nor techno, and don't care much about "air" around the instruments (exaggerated top octave?) nor the "soundstage" effects (wide dispersion?) nor "dry, fast" bass (dip in the upper-bass?) nor "enhanced" detail (mid-woofer breakup resonance?).

I'm sure that the BBC sound engineers were far more well equipped to determine what was/is the "correct" or "neutral" balance at the listening spot. Not only because they dealt with live and recorded sound on a daily basis but also because unlike speaker manufacturers they were able to compare the sound coming out of the speakers with the real thing (the BBC runs a number of ensembles including large choirs and symphony orchestras and broadcasts classical as well as other types of music).


As for directional characteristics, again I think it's easy to again make a case regarding the different requirements of distinct music genres.
Good recordings of unamplified music played in naturally reverberant venues allow the listener to be drawn into the original soundscape. Listening room boundary reflections will have a negative effect, interfering of even destroying the illusion of attending a musical event.
Studio produced music on the other hand may sometimes benefit from boundary reflections. Such recordings constist on a collage of mono tracks that close-mic'ed in a semi-anechoic space(s) and then manipulated to create the music. They sometimes sound "dry" and "uninvolving" and they're hardly ever meant to sound realistic.

Besides the boundaries of most domestic listening rooms are untreated and not always symmetrical. Side-wall reflections in smaller European or Asian sitting rooms will have significantly more impact than in the plentyful space in lounges of North American homes.
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
I'd really like to know of alternatives to the Olive studies on this question. As mentioned above, I have a lot of reservations about the Olive research, but I know of little alternative data.

Not sure if you've seen an old paper by Brüel & Kjær titled "Relevant loudspeaker tests in studios, in Hi-Fi dealers' demo rooms, in the home etc. using 1/3 octave, pink-weighted, random noise" which is in many ways identical to the Harman research.
One aspect worth noting is that those tests were performed in real domestic conditions.
 

raistlin65

Major Contributor
Forum Donor
Joined
Nov 13, 2019
Messages
2,279
Likes
3,421
Location
Grand Rapids, MI
IIUC, you're speculating that in the 2004 studies, Olive biased the speaker sample selection (either intentionally or unintentionally) so that all or most of the neutral speakers did not have other issues, and so that many or most of the non-neutral speakers did.

Did I understand you correctly?

I'm saying that I am skeptical of the reliability of that study. The methodology writeup seems flawed, based upon what you have quoted about sample selection. Could be it was addressed where he defines neutral vs. non-neutral in the text. But I'd need to read it.

And I'm saying, what I've always been saying, because of what Sean Olive wrote in those comments that I previously quoted to you. In 2015, it seems like he's walking back the previous research a little where he is responding to questions about the headphone tone control research.

Anyway, if you or someone has a link to the 2004 loudspeaker preference research that is not behind a paywall, I'd be glad to take a look at it. I really cannot talk about this any more than I already have without examining the article itself.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
It's an indisputable fact that the Harman data was collected in a scientific manner (evidence-based) but I don't see how "average listener preference" relates to accurate reproduction of the recorded signal. In my view it merely provides information regarding people's tastes.

I agree with this 100%. That's precisely what the study was designed to investigate.

Regarding frequency response, most people don't listen to recordings of acoustic music and some have never listened to a single live unamplified concert in their life; "realism" is not a requirement for the majority. Besides it is impossible to determine "realism" unless one is familiar with live and uses long term listening assessment and adequate recordings. An short-duration AB comparison is totally unfit for such purpose.

I agree with this also (apart from AB comparison being unfit). This was not what was being tested either. All that was being tested was preference.

As to the duration of the tests, what do you mean by "short"?

I'm sure that the BBC sound engineers were far more well equipped to determine what was/is the "correct" or "neutral" balance at the listening spot. Not only because they dealt with live and recorded sound on a daily basis but also because unlike speaker manufacturers they were able to compare the sound coming out of the speakers with the real thing (the BBC runs a number of ensembles including large choirs and symphony orchestras and broadcasts classical as well as other types of music).

I'm sure they were. I just don't see their aims as the same as those of Olive et al, who were interested in finding out what people preferred.

How did the recommendations of the BBC sound engineers differ from those of Olive?

As for directional characteristics, again I think it's easy to again make a case regarding the different requirements of distinct music genres.
Good recordings of unamplified music played in naturally reverberant venues allow the listener to be drawn into the original soundscape. Listening room boundary reflections will have a negative effect, interfering of even destroying the illusion of attending a musical event.
Studio produced music on the other hand may sometimes benefit from boundary reflections. Such recordings constist on a collage of mono tracks that close-mic'ed in a semi-anechoic space(s) and then manipulated to create the music. They sometimes sound "dry" and "uninvolving" and they're hardly ever meant to sound realistic.

Yep, this makes sense to me. But Olive's research didn't find that loudspeakers should have a particular beamwidth if they're to be preferred, but rather that its off-axis response should be smooth and should closely match the on-axis response. Wouldn't this hold regardless whether the speaker's beamwidth were 180° or 90°, for example?

Not sure if you've seen an old paper by Brüel & Kjær titled "Relevant loudspeaker tests in studios, in Hi-Fi dealers' demo rooms, in the home etc. using 1/3 octave, pink-weighted, random noise" which is in many ways identical to the Harman research.
One aspect worth noting is that those tests were performed in real domestic conditions.

Yes, I can see that they were interested in similar things. What do you see this study as saying about loudspeaker preferences/requirements?
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
And I'm saying, what I've always been saying, because of what Sean Olive wrote in those comments that I previously quoted to you. In 2015, it seems like he's walking back the previous research a little where he is responding to questions about the headphone tone control research.

Is he walking back from it in your opinion in the sense that he's admitting that preference is also correlated with experience and age?

Or is it something to do with his "admitting" that many of the speakers in the original study had other issues? (which I see as old news for the reasons we've discussed)
 

raistlin65

Major Contributor
Forum Donor
Joined
Nov 13, 2019
Messages
2,279
Likes
3,421
Location
Grand Rapids, MI
Is he walking back from it in your opinion in the sense that he's admitting that preference is also correlated with experience and age?

Or is it something to do with his "admitting" that many of the speakers in the original study had other issues? (which I see as old news for the reasons we've discussed)

We are going in circles, and I find that I've been replying to you with things I have said previously. Usually a good sign it's time to move on to other topics :)
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
Yep, this makes sense to me. But Olive's research didn't find that loudspeakers should have a particular beamwidth if they're to be preferred, but rather that its off-axis response should be smooth and should closely match the on-axis response. Wouldn't this hold regardless whether the speaker's beamwidth were 180° or 90°, for example?

Speakers with a widedispersion pattern interact more with room boundaries in the midrange and treble. In my experience this affects imaging negatively and sometimes tonal balance too, particularly when the first reflection zones aren't treated.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Speakers with a widedispersion pattern interact more with room boundaries in the midrange and treble. In my experience this affects imaging negatively and sometimes tonal balance too, particularly when the first reflection zones aren't treated.

Ok yes, that's true (although I believe it's a good thing in the horizontal plane).
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Would any of you who are saying that flat-measuring speakers are not the most preferred be willing to suggest how you believe speakers should measure if they're to be preferred?
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
Would any of you who are saying that flat-measuring speakers are not the most preferred be willing to suggest how you believe speakers should measure if they're to be preferred?

I think that the anechoic frequency response should be flat on-axis.

What I have been trying to propose is that optimal horizontal off-axis characteristics (at what frequency the speakers start beaming and by how much) are somewhat room dependent though I entirely agree that the response should be smooth off-axis.

Speakers which are flat on-axis can either have:

constant narrow directivity (usually horn-loaded),
709AGUfig5.jpg

Avantgarde Uno Nano

a flat-ish wide off-axis response (usually waveguided),
1113PSBT2fig5.jpg

PSB T2

a wide response with pronounced tilt in the treble (waveguided and/or wide baffle),
SFSfig5.jpg

Sonus Faber Stradivari

a dip in the crossover region followed by increasing directivity as you move away from the bottom of the tweeter's passband (no waveguide).
318harbeth.H302fig4.jpg

Harbeth M30.2
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I think that the anechoic frequency response should be flat on-axis.

What I have been trying to propose is that optimal horizontal off-axis characteristics (at what frequency the speakers start beaming and by how much) are somewhat room dependent though I entirely agree that the response should be smooth off-axis.

Ok yes, sure :)

What I don't understand is how this contradicts Olive's research?

Speakers which are flat on-axis can either have:

constant narrow directivity (usually horn-loaded),
709AGUfig5.jpg

Avantgarde Uno Nano

a flat-ish wide off-axis response (usually waveguided),
1113PSBT2fig5.jpg

PSB T2

a wide response with pronounced tilt in the treble (waveguided and/or wide baffle),
SFSfig5.jpg

Sonus Faber Stradivari

a dip in the crossover region followed by increasing directivity as you move away from the bottom of the tweeter's passband (no waveguide).
318harbeth.H302fig4.jpg

Harbeth M30.2

Actually, I'm not sure what you're trying to illustrate with these examples.

Let's take it as a given that some speakers have a smooth off-axis response (more similar to your first and second examples, notwithstanding that neither is quite ideal), and some speakers do not (e.g. your last example).

According to Olive, speakers of types 1 and 2 would be more likely to be preferred.

So again, I don't see where your opinion differs from Olive's?
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
According to Olive, speakers of types 1 and 2 would be more likely to be preferred.

So again, I don't see where your opinion differs from Olive's?

As I said, I disagree because I find that the "optimal horizontal off-axis characteristics" are somewhat room as well as listener dependent.

Type 1 interacts the least with the room, yet many people prefer type 2 becase it interacts the most with the room because it enhances the "spaciousness" or "soundstage" effect (Linkwitz talks about it here http://www.linkwitzlab.com/Recording/phantom images.pdf ).

Many people rate "soundstage" higher than "imaging" or tonal balance.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
As I said, I disagree because I find that the "optimal horizontal off-axis characteristics" are somewhat room as well as listener dependent.

Type 1 interacts the least with the room, yet many people prefer type 2 becase it interacts the most with the room because it enhances the "spaciousness" or "soundstage" effect (Linkwitz talks about it here http://www.linkwitzlab.com/Recording/phantom images.pdf ).

Many people rate "soundstage" higher than "imaging" or tonal balance.

Ok, I might be misunderstanding Olive or may have overlooked some aspect of the research. Where does he say that a particular horizontal off-axis characteristic (other than smooth) is preferable?
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
Ok, I might be misunderstanding Olive or may have overlooked some aspect of the research. Where does he say that a particular horizontal off-axis characteristic (other than smooth) is preferable?

I have no idea if he said it or not.
All I know is that Harman speakers (Revel, Infinty, budget JBL) have very wide dispersion characteristics, flat into the high-thousands and with very little attenuation as you move away from the axis. There's a fair chance that he is practicing what he preaches.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I have no idea if he said it or not.
All I know is that Harman speakers (Revel, Infinty, budget JBL) have very wide dispersion characteristics, flat into the high-thousands and with very little attenuation as you move away from the axis. There's a fair chance that he is practicing what he preaches.

No, he doesn't say it to my knowledge. You and he are in agreement as far as I understand it.

And Revel's professional series has a number of speakers that are narrower dispersion than their home audio speakers btw, although not as narrow as the Avantgarde, which is an extreme case.
 

TomB19

Active Member
Joined
Oct 13, 2019
Messages
137
Likes
102
I wonder what would happen if somebody were to implement loudness again but in an epic way with DSP processing.

Specifically, if you could sit down and work with your system on a sinusoidal frequency sweep to approximately balance the amplitude at all frequencies. Doing that at a few different levels would allow the system to adjust the signal for a specific user. As the user ages, they could email that profile to their hearing aid vendor. :D

It would be the ultimate indulgence in subjectivity but, by doing it in software, it would be easily defeatable and adjustable.

Something like this might just set off a wave of bell bottom trousers that hasn't been seen in 45 years.
 
Last edited:

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I wonder what would happen if somebody were to implement loudness again but in an epic way with DSP processing.

Specifically, if you could sit down and work with your system on a sinusoidal frequency sweep to balance the amplitude at all frequencies. Doing that at a few different levels would allow the system to adjust the signal for a specific user.

It would be the ultimate indulgence in subjectivity but, by doing it in software, it would be easily defeatable and adjustable.

Something like this might just set off a wave of bell bottom trousers that hasn't been seen in 45 years.

I've tried this using a dynamic equaliser run via Equaliser APO.

Takes some time to calibrate and requires some subjective input as there is no definitive set of equal loudness curves, but works pretty well
 

TomB19

Active Member
Joined
Oct 13, 2019
Messages
137
Likes
102
I've tried this using a dynamic equaliser run via Equaliser APO.

Takes some time to calibrate and requires some subjective input as there is no definitive set of equal loudness curves, but works pretty well

How did you define the amplitude adjusted equalization? It would be easy to configure a curve specific to one individual but I suspect the tricky bit would be making modifying the equalization curve based on amplitude, the way loudness does.

Does the APO have a facility for that? Perhaps an interface that allows the user to enter a Laplace transform? lol!
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
How did you define the amplitude adjusted equalization? It would be easy to configure a curve specific to one individual but I suspect the tricky bit would be making modifying the equalization curve based on amplitude, the way loudness does.

Does the APO have a facility for that? Perhaps an interface that allows the user to enter a Laplace transform? lol!

That's what the dynamic EQ is for. You set up an EQ curve corresponding to the difference between the equal loudness curve at the lowest level you listen at and (IMO) the equal loudness curve at 80dB, which is just shy of the level at which engineers are supposed to mix/master at.

Then you set the threshold and attack such that, whenever the level drops below 80dB, the EQ curve begins to take effect, with the magnitude of its effect ranging from 0% at and above 80dB to 100% at whatever level you specified as your lowest level in the previous step.

Ofc, your volume control gas to come before the dynamic EQ for it to work correctly!

There may be a more elegant way to do it, I got distracted by other things in the end and didn't persevere.
 
Top Bottom