Anyone ever read Harry Olson's Elements of acoustical engineering?

[richard12511]

I was watching this interview with Jonathan Weiss of Oswalds Mill Audio. In the video, he makes several (imo)controversial statements(ex: horns should be straight, not curved), but the most interesting of those is that speakers should be designed with a lowercase n - or inverse v - curve. He also says the tradition Toole view that speakers should be neutral is incorrect, as it conflicts with the way humans hear. I tend to mostly ignore statements like this as marketing, but what made this statement more compelling - at least to me - was that he actually cited a source, "Elements of acoustical engineering" - Harry Olson. I haven't read that book, but I'm wondering if anyone here has. If you have, is what he's saying a correct interpretation of Olson? and if so, does it have merit?

Here is the timestamped version of the video posted above.

 

[pozz]

Hard not to just dismiss him.

Conical horns are exceptionally coloured and have very uneven dispersion. I don't have a digital version, but the recent book on horns by Kolbrek and Dunker show measurements of this behaviour.

He quotes Olson but explains very little of the reasoning. That FR with shallow slopes on either end could be from pg. 601 (from the section called Frequency Range Preference for Reproduced Speech and Music) of Olson's book (below). Olson states:

As contrasted to the other data presented, the purpose of this investigation was the determination of the frequency range of reproduced speech and music that is most pleasant to the average listener. ... The general conclusion of these tests is that listeners prefer either a narrow or medium frequency range to a wide one. However, the exact choice of bandwidth varies to some extent within these limits for differen types of program content. Listeners prefer a narrow to a wide tonal range even when informed that one condition is low fidelity and the other high fidelity. ... It is interesting to note that the preferred frequency range of a representative cross section of broadcast listeners is essentially the same as the frequency range of commercial radio receivers, phonographs, sound motion picture systems, and sound systems.

However, Weiss seems to have stopped reading at this point. Olson says the following:

There are three possible reasons for the results of these tests, as follows:

1. The average listener, after years of listening to the radio and the phonograph, has become conditioned to a restricted frequency range and feels that this is the natural state of affairs.
2. Musical instrumenets are not properly designed and would be more pleasing and acceptable is the production of fundamentals and overtones in the high-frequency range were suppressed. [A little explanation is needed here: Olson posted a chart showing the frequency ranges of popular instruments and male and female voices. The ranges far exceed the "preferred" FR.]
3. The distortions and deviations from true reproduction of the original sound are less objectionale with a restricted frequency range.

Olson then goes on to produce the following list:

#1 being most important. Olson then goes on to describe a huge blind-listening experiment where they built a system to reproduce a six-person orchestra. Among its characteristics is very wideband and flat on-axis FR:

And Olson is no fool about directivity:

Or distortion:

Or correct SPL for the flattest subjective loudness:

And the results?

So that's that.

(By the way. Olson is one of Toole's favourite citations for both loudspeaker design and blind testing.)

Finally, time-domain "accuracy" or "smearing" have never been shown to be a prime factor in loudspeaker design. There is a good paper by Vanderkooy about this: https://www.aes.org/e-lib/browse.cfm?elib=3824


__________________________________

Man. The time it takes to unwind a bunch of BS.

 

[richard12511]

Hard not to just dismiss him.

Conical horns are exceptionally coloured and have very uneven dispersion. I don't have a digital version, but the recent book on horns by Kolbrek and Dunker show measurements of this behaviour.

He quotes Olson but explains very little of the reasoning. That FR with shallow slopes on either end could be from pg. 601 (from the section called Frequency Range Preference for Reproduced Speech and Music) of Olson's book (below). Olson states:


View attachment 93059
However, Weiss seems to have stopped reading at this point. Olson says the following:

Olson then goes on to produce the following list:
View attachment 93061
#1 being most important. Olson then goes on to describe a huge blind-listening experiment where they built a system to reproduce a six-person orchestra. Among its characteristics is very wideband and flat on-axis FR:
View attachment 93063
And Olson is no fool about directivity:
View attachment 93071

Or distortion:
View attachment 93065

Or correct SPL for the flattest subjective loudness:
View attachment 93070

And the results?
View attachment 93072

So that's that.

(By the way. Olson is one of Toole's favourite citations for both loudspeaker design and blind testing.)

Finally, time-domain "accuracy" or "smearing" have never been shown to be a prime factor in loudspeaker design. There is a good paper by Vanderkooy about this: https://www.aes.org/e-lib/browse.cfm?elib=3824


__________________________________

Man. The time it takes to unwind a bunch of BS.

Thanks for the very detailed response. Kinda what I figured.

I knew that Toole often cites Olson, which is a big part of why (having never read Olson directly) I couldn't just dismiss this guy right away.

 

[andreasmaaan]

Hard not to just dismiss him.

Conical horns are exceptionally coloured and have very uneven dispersion. I don't have a digital version, but the recent book on horns by Kolbrek and Dunker show measurements of this behaviour.

He quotes Olson but explains very little of the reasoning. That FR with shallow slopes on either end could be from pg. 601 (from the section called Frequency Range Preference for Reproduced Speech and Music) of Olson's book (below). Olson states:


View attachment 93059
However, Weiss seems to have stopped reading at this point. Olson says the following:

Olson then goes on to produce the following list:
View attachment 93061
#1 being most important. Olson then goes on to describe a huge blind-listening experiment where they built a system to reproduce a six-person orchestra. Among its characteristics is very wideband and flat on-axis FR:
View attachment 93063
And Olson is no fool about directivity:
View attachment 93071

Or distortion:
View attachment 93065

Or correct SPL for the flattest subjective loudness:
View attachment 93070

And the results?
View attachment 93072

So that's that.

(By the way. Olson is one of Toole's favourite citations for both loudspeaker design and blind testing.)

Finally, time-domain "accuracy" or "smearing" have never been shown to be a prime factor in loudspeaker design. There is a good paper by Vanderkooy about this: https://www.aes.org/e-lib/browse.cfm?elib=3824


__________________________________

Man. The time it takes to unwind a bunch of BS.

 

[pozz]

Thanks for the very detailed response. Kinda what I figured.

I knew that Toole often cites Olson, which is a big part of why (having never read Olson directly) I couldn't just dismiss this guy right away.

It was a fun exercise. Olson's book really is a classic.

 

[tuga]

Hard not to just dismiss him.

This.

OMA = Good carpentry skills and great salesmanship. For hi-fi look elsewhere.

 

[andreasmaaan]

This.

OMA = Good carpentry skills and great salesmanship. For hi-fi look elsewhere.

Oh it's this company!

 

[Speedskater]

I read Olson's "Modern Sound Reproduction" book several decades ago. Don't remember much about it.

 

[pozz]

Oh it's this company!

When they were featured by Bloomberg a while back it was ... very weird. I don't know what makes them or their products newsworthy.

 

[andreasmaaan]

When they were featured by Bloomberg a while back it was ... very weird. I don't know what makes them or their products newsworthy.

Fortunately (as I could already feel the sick rising up the back of my throat) the majority of that article is behind a paywall

 

[richard12511]

In his blog, he explains his reasoning for using conical horns.

I actually don’t have a problem with audiophiles telling me how much they hate the sound of horns, because every single high end company making horn loudspeakers makes them completely differently from OMA. If you look at any other horn, you’ll see the walls are curved. This profile can and usually is of the exponential flare, there are also parabolic, hyperbolic and other flares, and in the exponential category there are sub species such as tractrix, spherical, LeCleach, etc. They are all similar in that their walls have curvature. That may not seem like such a big deal, but it is.

Conical horns are completely different. They have straight sided walls. The sound comes out of the driver at the throat end and sees a perfectly consistent expansion rate, which means the sound wave emerges as a spherical wavefront without any perturbations. That’s not what happens with a curved horn, because the sound wave is now seeing curved walls, a bit like playing pool on a curved billiard table. As the frequency goes up, the sound waves don’t even “see” the horn walls and start to beam, which is why you’ll often see conventional horn speakers toed in to aim at a central listening chair or position- if you’re not sitting there with your head in an imaginary vise you will miss part of the essential high frequency information because its only to be heard on axis. This makes a listening session at a horn afficianado’s home frustrating, because only one person at a time can really hear the music properly.

Conical horns don’t have any of these problems. In fact, one of their attributes is a technical term called “constant directivity” which is one of the absolute most desirable characteristics in the pro audio world. What it really means is that everywhere the horn is aiming (defined by looking at the angles of the horn, and pretending that if it were a fire hose nozzle, who would get soaked sitting in front of it) gets to hear the same thing in the same way.

Many pro horns have been created to have constant directivity through complex throat and mouth geometries, but conicals have this naturally. The other thing they have is a totally natural sound, without colorations typical of other curved horns. This is the real sticking point to my mind, why audiophiles have issues with horn speakers. The problem is the constriction at the throat typical of these horns. Its kind of like the difference between a trumpet and a megaphone. A trumpet is loud, really loud, and its quite short. A megaphone is not so loud, but a voice coming through it is clear. The trumpet is much louder because its more efficient, specifically it couples the air in the mouth of the trumpeter to the surrounding air better than the megaphone, because that constriction presents a small air column which rapidly expands to the sound waves. The conical horn does not have that constriction.

I don't agree with his reasoning(it seems counter to everything I've learned so far), but he does at least explain it. Plus, that Monarch speaker does look pretty cool.

 

[richard12511]

When they were featured by Bloomberg a while back it was ... very weird. I don't know what makes them or their products newsworthy.

Really well made video. I could definitely see where a super busy billionaire that doesn't have time to understand audio science could get suckered into buying one of these systems.

 

[pozz]

In his blog, he explains his reasoning for using conical horns.



I don't agree with his reasoning(it seems counter to everything I've learned so far), but he does at least explain it. Plus, that Monarch speaker does look pretty cool.

It's really unclear where he's getting his information from. He's so upfront and so wrong at the same time.

 

[richard12511]

It's really unclear where he's getting his information from. He's so upfront and so wrong at the same time.

Uber high end audio is crazy

 

[andreasmaaan]

It's really unclear where he's getting his information from. He's so upfront and so wrong at the same time.

The only 100% error-free substantive statement I can find in there concerning conical horns is:

Conical horns... have straight sided walls.

Can't argue with that

 

[Tom C]

It seems like the designer here had one primary goal: how to produce something for which an exceptionally high price can be charged. Not, how to make something that sounds especially good, or how to improve performance in any way. Just, how to make the sale price high.