Evidence-based Speaker Designs

[thewas]

I tought we were talking about measurements at the LP position?

Yes, as I wrote above the slope at the LP doesn't necessarily describe the perceived tonality above transition frequency.

 

[bigjacko]

The recommended setup for Earl's constant-directivity waveguide speakers, and for mine (most of which use a non-axisymmetric device), is to use aggressive toe-in such that the speaker axes criss-cross in front of the listening area, and I can explain the reasoning if you would like.

the hole is a result of a very coherent wavefront emerging from the throat. That a different horn doesn't have such a null because the wavefront isn't as coherent doesn't necessarily mean the horn is superior.

Thanks for sharing those informations, can you guys also tell a bit more on why coherent wavefront produce a hole on axis and why aggressive toe in is better?

 

[andreasmaaan]

that’s my experience as well. Still if all studios are using a flat target, you can learn to mix for that and it will be consistent. if people start to use a tilt provided by Sonarworks or similar but they do not agree on the slope we will not improve.

Yes, but as @thewas mentioned, the problem is that our ears/brain don't derive our perception of a system's tonality simply from the in-room steady-state response. Instead, we tend to derive tonality primarily from a speaker's direct sound at frequencies whose wavelengths are short relative to the room on one hand, and from the combined sound field at lower frequencies whose wavelengths are long relative to the room on the other.

For this reason, a flat in-room response will not have the same perceived tonality in Room A as it does in Room B (especially in the all-important midrange and treble).

When I said I found the uptake in use of target-curve based systems like Sonarworks to be a concerning development, it was for this reason. And that goes for systems that EQ to a flat in-room response and systems that EQ to some other (e.g. Harman-like) target; both are based on the same fallacy.

 

[pierre]

Yes, but as @thewas mentioned, the problem is that our ears/brain don't hear don't derive our perception of a system's tonality simply from the in-room steady-state. Instead, we tend to derive our perception of a speaker's tonality primarily from its direct sound at frequencies whose wavelengths are short relative to the room on one hand, and from the combined sound field at lower frequencies whose wavelengths are long relative to the room on the other.

For this reason, a flat in-room response will not have the same perceived tonality in Room A as it does in Room B (especially in the all-important midrange and treble).

When I said I found the uptake in use of target-curve based systems like Sonarworks to be a concerning development, it was for this reason. And that goes for systems that EQ to a flat in-room response and systems that EQ to some other (e.g. Harman-like) target; both are based on the same fallacy.

The graph from Genelec is for near field or far field? Near field, it should not matter much, reflections are usually very attenuated in studio (less so at home). Far field, for mastering, I agree. Mastering room are usually dead. Is controlling the RT60 enough to get consistent perceived sound?

 

[andreasmaaan]

The graph from Genelec is for near field or far field? Near field, it should not matter much, reflections are usually very attenuated in studio (less so at home). Far field, for mastering, I agree. Mastering room are usually dead. Is controlling the RT60 enough to get consistent perceived sound?

I'm not sure which Genelec graph you mean?

But yeh, that's a very good point. For less reverberant rooms and for nearfield listening, the difference between the in-room response and the direct sound is going to be much less significant than for more reverberant rooms with the listener in the farfield. So EQing to a target curve will be less of a problem in control rooms than in, say, a typical living room.

 

[thewas]

I'm not sure which Genelec graph you mean?

I think he means this one this one which was taken at the LP of many studios, although we don't know at which distances that was and I think rather quite varying which would also explain the wide tolerance band.

 

[dominikz]

I'm not sure which Genelec graph you mean?

But yeh, that's a very good point. For less reverberant rooms and for nearfield listening, the difference between the in-room response and the direct sound is going to be much less significant than for more reverberant rooms with the listener in the farfield. So EQing to a target curve will be less of a problem in control rooms than in, say, a typical living room.

Probably a redundant comment, but just wanted to say that this has also been my experience with (automatic) room correction.

It seems to me EQ-ing to a flat target in the nearfield can work well, but in the far-field one typically has to adapt the slope or even the whole shape of the target curve to the measured in-room response to get 'flat' tonality. I agree it is unfortunate that default target curves in such SW are typically horizontal lines as that is bound to mislead many users.
Guess the optimal way to approach in-room EQ would be to use room correction below Schroeder only, and if required do speaker EQ based on anechoic LW measurement above it.

 

[ernestcarl]

Guess the optimal way to approach in-room EQ would be to use room correction below Schroeder only, and if required do speaker EQ based on anechoic LW measurement above it.

Even without anechoic measurements, if you use adequate spatial averaging (like MMM), you can make small corrections well above the transition zone if really needed.

 

[Ilkless]

Thanks for sharing those informations, can you guys also tell a bit more on why coherent wavefront produce a hole on axis and why aggressive toe in is better?

Replies here and elsewhere have made me update my understanding of the hole on-axis issue. The subject of this comment (Geddes OS waveguide) is indeed an exceptionally-low distortion waveguide geometry overall. It may throw anomalies that would typically be drowned out in lesser designs into sharp relief. The hole is caused by a sub-optimal mouth termination due to packaging and manufacturing constraints, something that is compounded by the symmetry of the waveguide. Some sound bounces back across the mouth; as the distance from the centre of the waveguide is the same to all points along the mouth, the sound bounces back with an on-axis interference. But this is splitting hairs of course. The tonal balance of direct/indirect sound is determined by radiation along a range of angles, not a single angle, so a null singularly on-axis can have a benign overall effect.

Aggressive toe-in is better for speakers with controlled, smooth directivity because it can broaden the sweet spot. See here.

 

[Duke]

Thanks for sharing those informations, can you guys also tell a bit more on why coherent wavefront produce a hole on axis and why aggressive toe in is better?

Ilkless explained it well.

Saying the same thing with different words: There will be a reflection around the mouth of most horns, unless the round-over is quite large. With a round horn, that reflection will all arrive on-axis at the same time (whereas with a rectangular or elliptical horn, the reflection arrives spread out in time somewhat - NOT coherent). So for a given on-axis listening distance, there will be frequency at which the mouth reflection arrives 1/2 wavelength later than the straight-down-the-middle-of-the-horn sound. And so in the general region of that frequency, you'll have some dippage. Typically the dip will be centered somewhere in between 5 kHz and 10 kHz, but it depends on the specific path lengths involved.

The solution Geddes recommends is to listen from about 15-20 degrees off-axis, where the mouth reflection is no longer coherent, and the dip disappears.

On the recommendation for aggressive toe-in with this type of loudspeaker, the main benefit is a much wider sweet spot. Briefly, the ear localizes sound by two mechanisms: Arrival time and intensity. With the aggressive toe-in, for an off-centerline listener, the near speaker "wins" arrival time but the far speaker "wins" intensity, because the listener is essentially on-axis of the far speaker but well off-axis of the near speaker. The net result is a pretty good soundstage even from well off to either side. The SECRET to this working well is, the output of that near speaker must fall off smoothly and rapidly as you move off-axis.

Take a look at the first photo here and you'll see what I mean. (Disclaimer: It's a photo of one of my systems at an audio show.)

 

[Kvalsvoll]

Yes, but as @thewas mentioned, the problem is that our ears/brain don't derive our perception of a system's tonality simply from the in-room steady-state response. Instead, we tend to derive tonality primarily from a speaker's direct sound at frequencies whose wavelengths are short relative to the room on one hand, and from the combined sound field at lower frequencies whose wavelengths are long relative to the room on the other.

For this reason, a flat in-room response will not have the same perceived tonality in Room A as it does in Room B (especially in the all-important midrange and treble).

When I said I found the uptake in use of target-curve based systems like Sonarworks to be a concerning development, it was for this reason. And that goes for systems that EQ to a flat in-room response and systems that EQ to some other (e.g. Harman-like) target; both are based on the same fallacy.

This is very important to understand, it may be obvious to many experts, but perhaps still not communicated well enough to have broad acceptance.

The same system in different rooms can have very different tilt on the frequency response, and it will not sound right and they will not sound similar in tonal balance if you choose to equalize those to the same target response.

I have now acquired measurements of the same system in different rooms with very different acoustic properties. What I see is the the tilt varies from around +5dB/close to flat in a properly treated quite dry room, +8dB in a untreated lager space, +12dB in a typical untreated domestic living room. Whether they sound "right" or good or bad can surely be debated, but it is quite clear that equalizing those to the same target will destroy the tonal balance in 2 of them. The flat room does not sound bright compared to the rooms with seemingly more bass in the measurements.

It is also interesting to observe that the response follows a gradual tilt so that the more live rooms gain more spl at low frequencies, but also a smooth gradually decreasing increase upwards in frequency.

 

[williamwally]

...I have now acquired measurements of the same system in different rooms with very different acoustic properties. What I see is the the tilt varies from around +5dB/close to flat in a properly treated quite dry room, +8dB in a untreated lager space, +12dB in a typical untreated domestic living room...

Would you mind posting more/making a new thread with this info?

 

[Kvalsvoll]

Would you mind posting more/making a new thread with this info?

There are 2 parts of this - the system consisting of speakers and room, which creates the measured response, and then there is the perception part, which is how we hear the sound.

The first part is not very difficult, and relates to this thread. The other part about perception, is difficult. I have no good description for that, and to make something that holds up scientifically - which I consider is a requirement if it is to have any value - needs more works and likely some controlled experiments. Perhaps not so suitable for a forum.

I have written briefly on this subject in articles and on my own forum page, but the text only covers the speaker-room part and essentially says nothing more than my previous post in this thread.

The acoustic properties of a typical untreated room gives a fairly flat frequency response of decay - all frequencies decays at similar rate.

A speaker with a tilted power response such that the radiation narrows from very wide at low frequencies into quite narrow at high frequencies will then give a tilted measured frequency response. If the speaker has huge steps or other large deviations from a smooth curve in off-axis response, this will show up in the measured in-room response as deviations from a smooth, tiltet curve.

In a properly treated room, the frequency response of the decay follows a similar flat curve, it just decays much faster. The acoustic treatment is carefully designed to preserve energy at higher frequencies. Rooms with too little low frequency absorption and too much higher frequency absorption seems to be how many treated rooms ended up, resulting in a dull sound with too little high frequency energy.

The reason we measure more flat in a treated room is that more of the speakers on-axis response dominates, but the reflected energy in the room is similar in spectral balance. It is just that the levels of this sound is too low to make a large enough contribution to affect the frequency response.

 

[q3cpma]

Ilkless explained it well.

Saying the same thing with different words: There will be a reflection around the mouth of most horns, unless the round-over is quite large. With a round horn, that reflection will all arrive on-axis at the same time (whereas with a rectangular or elliptical horn, the reflection arrives spread out in time somewhat - NOT coherent). So for a given on-axis listening distance, there will be frequency at which the mouth reflection arrives 1/2 wavelength later than the straight-down-the-middle-of-the-horn sound. And so in the general region of that frequency, you'll have some dippage. Typically the dip will be centered somewhere in between 5 kHz and 10 kHz, but it depends on the specific path lengths involved.

The solution Geddes recommends is to listen from about 15-20 degrees off-axis, where the mouth reflection is no longer coherent, and the dip disappears.

On the recommendation for aggressive toe-in with this type of loudspeaker, the main benefit is a much wider sweet spot. Briefly, the ear localizes sound by two mechanisms: Arrival time and intensity. With the aggressive toe-in, for an off-centerline listener, the near speaker "wins" arrival time but the far speaker "wins" intensity, because the listener is essentially on-axis of the far speaker but well off-axis of the near speaker. The net result is a pretty good soundstage even from well off to either side. The SECRET to this working well is, the output of that near speaker must fall off smoothly and rapidly as you move off-axis.

Take a look at the first photo here and you'll see what I mean. (Disclaimer: It's a photo of one of my systems at an audio show.)

Wonder why companies who have the means of doing the perfect (with a minimum of throat exit reflection) OS waveguide didn't do so and went with designs nearer to straigth or exponential slopes. I'm thinking anybody casting or moulding their front baffle (Genelec, Neumann, D&D, etc...).

 

[Scgorg]

Wonder why companies who have the means of doing the perfect (with a minimum of throat exit reflection) OS waveguide didn't do so and went with designs nearer to straigth or exponential slopes. I'm thinking anybody casting or moulding their front baffle (Genelec, Neumann, D&D, etc...).

To my knowledge the OS waveguide is designed for flat wavefronts, meaning a theoretically ideal compression driver. Direct radiating domes have spherical wave expansion, in which case a conical horn should be used (IIRC, of course). Neumanns waveguides look pretty much perfectly conical in expansion, which should be the "ideal" shape for the spherical wavefronts of their dome tweeters.

I think Geddes has mentioned on DIYaudio that the genelec S360 waveguide is strongly resembling an OS waveguide? Don't quote me on that though.

 

[Duke]

Wonder why companies who have the means of doing the perfect (with a minimum of throat exit reflection) OS waveguide didn't do so and went with designs nearer to straigth or exponential slopes. I'm thinking anybody casting or moulding their front baffle (Genelec, Neumann, D&D, etc...).

It's always a juggling of tradeoffs. The Dutch & Dutch design was informed by Earl's work on the Oblate Spheroid, so there is some family resemblance. My guess is that adaptations were made to accommodate the non-flat wavefront of the dome tweeter, and to minimize the on-axis dip which has been mentioned. It looks to me like there might be a mild on-axis dip between 11 kHz and 14 kHz in some of the data, and that dip would move north as the listening distance increases, with its practical effect being correspondingly reduced.

The Genelec S360's waveguide looks like it could be an Elliptical Oblate Spheroid to me, which would have the advantage of having little or no on-axis dip in the response.

DIY Sound Group makes Super Elliptical Oblate Spheroid waveguides, which I've been using for years in some of my designs. Again, no on-axis dip.

 

[q3cpma]

It's always a juggling of tradeoffs. The Dutch & Dutch design was informed by Earl's work on the Oblate Spheroid, so there is some family resemblance. My guess is that adaptations were made to accommodate the non-flat wavefront of the dome tweeter, and to minimize the on-axis dip which has been mentioned. It looks to me like there might be a mild on-axis dip between 11 kHz and 14 kHz in some of the data, and that dip would move north as the listening distance increases, with its practical effect being correspondingly reduced.

The Genelec S360's waveguide looks like it could be an Elliptical Oblate Spheroid to me, which would have the advantage of having little or no on-axis dip in the response.

DIY Sound Group makes Super Elliptical Oblate Spheroid waveguides, which I've been using for years in some of my designs. Again, no on-axis dip.

By the way, since you seem to know a bit about the subject, do you have the answer to the question implied by this (https://www.audiosciencereview.com/...o-science-innovators.15025/page-2#post-472407) earlier post I made? Basically: were Genelec inspired by Geddes, or was the modern directivity matching waveguide developed in parallel and independently?

 

[Duke]

By the way, since you seem to know a bit about the subject, do you have the answer to the question implied by this (https://www.audiosciencereview.com/...o-science-innovators.15025/page-2#post-472407) earlier post I made? Basically: were Genelec inspired by Geddes, or was the modern directivity matching waveguide developed in parallel and independently?

I don't know anything about the history of Genelec's waveguides, sorry! My guess is "parallel and independently", at least at first.

 

[ElNino]

By the way, since you seem to know a bit about the subject, do you have the answer to the question implied by this (https://www.audiosciencereview.com/...o-science-innovators.15025/page-2#post-472407) earlier post I made? Basically: were Genelec inspired by Geddes, or was the modern directivity matching waveguide developed in parallel and independently?

I think the Genelec 1022A came out just a year after Geddes finished his Ph.D. and a few years prior to his first waveguide paper, so I'd guess Ilpo probably wasn't directly inspired by Geddes' work at the very beginning, but I'm sure he read Geddes 1987 paper when it came out.

 

[Senior NEET Engineer]

"Most acoustically advanced audiophile loudspeaker available on earth today"

https://www.tektondesign.com/ulfberht-pmd-monitor.html