I cannot trust the Harman speaker preference score

[Sean Olive]

I wish I had good news for you, but the original data file for these experiments disappeared in the move to California. There were several filing cabinets of data that had to be sorted through and either discarded or saved. Mistakes were made. I too would like to have another look at the original data. It is interesting that in almost 40 years nobody else has bothered to repeat or extend this kind of work and publish the results.

As for the choice of loudspeakers for the multiple-room tests, if you look at the authors of the paper my name is not among them. I had the facility for the experiment constructed while I was at the NRC directing the research, but the experiments were designed and run after I left to join Harman International. I agree with the choice of loudspeakers, though, because it was a difficult discrimination test and it truly showed the remarkable room adaptation that humans are capable of. The experiment, as you can imagine, was arduous and time consuming ($$$$), so it is understandable that it was not repeated using loudspeakers that would have been easier to recognize, such as those having distinctive directivities. In that respect, I have to say that at that time it would have been difficult to find examples having distinctive directivity that were similarly neutral - thereby allowing directivity to be the dominant variable. Most directionally distinctive loudspeakers of the time also were distinctive in their sound signatures, making them easily recognizable and invalidating the experiment. It could be done better these days, but finding someone with the competence and facilities to do it is a real challenge!

Personally, now that we have truly excellent "conventional" loudspeakers, and the option of supplemental surround/multichannel loudspeakers and processors I do think we have moved on. These days there is little or nothing to be gained in distinctive transduction processes beyond marketing claims. Loudspeakers with bipole, dipole, and omnidirectional behavior are not involved with the creation of the art in the professional audio domain, so their involvement in playback is that of adding a "sound effect" that people find pleasurable - it all comes down to attempts to extract more satisfaction from stereo. Whether it is real or imaginary is unimportant. I had monster bipole loudspeakers in one of my systems, as discussed and shown in my book. It was done deliberately for a special listening room (Figure 7.19). They would not be optimum for my existing listening circumstances (Figure 7.21) - references to Figures in the 3rd edition of my book.

Good point about conventional directivity (vs monopole, dipole, bipole, etc) of home speakers better match those used to make the recordings.. I forgot to mention that.

 

[sarumbear]

It is interesting that in almost 40 years nobody else has bothered to repeat or extend this kind of work and publish the results.

Interesting indeed…

 

[hardisj]

Mistakes were made.

Story of my life.

Glad to see you chiming in here, Floyd.

- Erin

 

[fineMen]

I still advertise to individually equalise the stereo setup @home to the standard for starters.

Means, make it generate a sound field, that within some margin matches the now clearly suggested Harman's "home curve".

On my own, non-commercial, behalf I would like to mention a possibility I found feasible for myself at my home, given my personal usage of the stereo.

Two not too directional speakers like 90° x 60° from 400Hz on, otherwise characterised by exceptionally low distortion and huge sound pressure level capability for the stereo panorama.
Accomplished by a single center speaker of wide dipole directivity, oriented 'on the edge', fed with a summed high passed mono signal, similar spl capabilities.
I've got several stored settings on my DSP to adjust the center speaker level and timbre to the current demand. Walking around cleaning the flat, socialising, checking the news, evaluating the detailed aesthetics of ancient recordings, following new musical ideas ...

The crux is, I'm left with nothing else to be desired. I've even miss out my daily 'critical listening' work-out.

 

[Multicore]

Late to the party, but here anyway . . .

My favorite sentence is...

Devotees of soundstage and imaging believe that fundamental issues are being overlooked.

Why's that? I am the anti-stereophile. I think stereo is not just greatly over-rated or at best a silly gimmick, I believe it is evil and should be denounced from the pulpit.

 

[MattHooper]

Why's that? I am the anti-stereophile. I think stereo is not just greatly over-rated or at best a silly gimmick, I believe it is evil and should be denounced from the pulpit.

Was that last comment serious?

Hard to tell on the interwebs.

 

[Newman]

Why's that?

Because the critics of the mono listening test think that it overlooks how well the speaker will do all the phantom imaging stuff between the speakers, eg height effects, depth effects, wide or narrow sound staging, the realism of the size of instruments between the speakers...that sort of thing.

 

[Floyd Toole]

Because the critics of the mono listening test think that it overlooks how well the speaker will do all the phantom imaging stuff between the speakers, eg height effects, depth effects, wide or narrow sound staging, the realism of the size of instruments between the speakers...that sort of thing.

The brain is a marvelous instrument - ask and it will deliver . . . and who can possibly argue?

 

[CtheArgie]

The brain is a marvelous instrument - ask and it will deliver . . . and who can possibly argue?

Woody Allen used to say that his brain was his second favorite organ….

 

[Pdxwayne]

The brain is a marvelous instrument - ask and it will deliver . . . and who can possibly argue?

I just asked my brain to make me a billionaire....Hope it will deliver soon.

; P

 

[Newman]

I had monster bipole loudspeakers in one of my systems, as discussed and shown in my book. It was done deliberately for a special listening room (Figure 7.19). They would not be optimum for my existing listening circumstances (Figure 7.21) - references to Figures in the 3rd edition of my book.

Hi Floyd, I am glad you mentioned this. Your fond recollection, and purchase for own use, of the M1 Mirage was used in another thread here to suggest you find omni speakers to be as good as any, perhaps even better ie Floyd Toole's choice from among all the speakers tested at the NRC in Canada, for use in his own home, (the same linked post even suggests that you are modifying the sound of 2-channel from your Revel Salon2's in your current setup via upmixing to capture some more of what the M1 did for you that the Revels can't), and prop up an argument that omni speakers are the ultimate way to reproduce sound, to wit, "Omnidirectional loudspeakers produce the most realistic musical soundstage in the home.... The superiority of the design is easily heard when in person, and when measured, particularly in the off-axis domains both vertically and horizontally. Live music and therefore sound propogation happens omnidirectionally, and is how our ear-brain mechanisms have evolved to understand sound. 98% of loudspeakers are designed incorrectly...(by not being omni)".

And when I tried to state that you used the M1 as a special case solution to a 'special case' room, and that they are suboptimal in most cases, the ensuing to-and-fro focused more on whether I had exaggerated your room as "virtually a giant echo chamber that just plain sounded bad" and the recordings of the day as "too-often horrendously poorly recorded and placed large sections of the orchestra wholly in one speaker, he needed a loudspeaker that was really bad at soundstage reproduction and greatly blurred the sonic image", LOL, maybe I'm guilty as charged there, but it didn't negate my main point.

So, it is interesting the many different ways you get referenced, in support of both sides of a discussion!

PS Also I suggested that sound sources are not omnidirectional, but it didn't seem to deflect the pro-omni-speaker lobby.

cheers

 

[Floyd Toole]

Hi Floyd, I am glad you mentioned this. Your fond recollection, and purchase for own use, of the M1 Mirage was used in another thread here to suggest you find omni speakers to be as good as any, perhaps even better ie Floyd Toole's choice from among all the speakers tested at the NRC in Canada, for use in his own home, (the same linked post even suggests that you are modifying the sound of 2-channel from your Revel Salon2's in your current setup via upmixing to capture some more of what the M1 did for you that the Revels can't), and prop up an argument that omni speakers are the ultimate way to reproduce sound, to wit, "Omnidirectional loudspeakers produce the most realistic musical soundstage in the home.... The superiority of the design is easily heard when in person, and when measured, particularly in the off-axis domains both vertically and horizontally. Live music and therefore sound propogation happens omnidirectionally, and is how our ear-brain mechanisms have evolved to understand sound. 98% of loudspeakers are designed incorrectly...(by not being omni)".

And when I tried to state that you used the M1 as a special case solution to a 'special case' room, and that they are suboptimal in most cases, the ensuing to-and-fro focused more on whether I had exaggerated your room as "virtually a giant echo chamber that just plain sounded bad" and the recordings of the day as "too-often horrendously poorly recorded and placed large sections of the orchestra wholly in one speaker, he needed a loudspeaker that was really bad at soundstage reproduction and greatly blurred the sonic image", LOL, maybe I'm guilty as charged there, but it didn't negate my main point.

So, it is interesting the many different ways you get referenced, in support of both sides of a discussion!

PS Also I suggested that sound sources are not omnidirectional, but it didn't seem to deflect the pro-omni-speaker lobby.

cheers

I hate it when people speak for me when they don't understand what I think! Omni speakers are playback manipulations, and if you like them, fine. If not, also fine. As for the notion that live music "happens omnidirectionally," it is simply not, NOT, true. Here is a Figure 10.15 from the 3rd edition.

 

[Kvalsvoll]

Late to the party, but here anyway . . .

As you might have imagined, I have been lurking and reading this somewhat confused, but interesting, thread for some time. I told Sean Olive about it, and he has jumped in with numerous substantive comments. He has dealt with many of the issues I felt compelled to respond to, but the subjective/objective correlation work was his and the initial challenge was to the rating itself. So, he is the first-responder authority on the topic.

This somewhat rambling response is based on writings that may eventually make their way to the companion website for my book (www.routledge.com/cw/toole). For the present purpose the document has grown overly long, so I will attach it as a pdf file for download. It is intended to expand on some of the discussion topics from perspectives that I thought were missed. I will refer to figures in my current book: “Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms”, 3rd edition, Focal Press 2017.

It took several coffee-breaks to get through this text. And though I disagree with some parts, I find there is much more common ground. To go into detail is beyond this random forum discussion.

So instead I make my comment on something we do agree upon. With a little addition at the end.

Flat on-axis frequency response looks like a no-brainer, and it is not very difficult to achieve. Yet many speakers show quite strange responses, many of which are highly rewarded. Are we missing something here? I think not. Some speakers are tuned to give a reasonably neutral tonal balance in-room, and if the off-axis is colored, the on axis needs to be adjusted. Then you end up with a speaker that can be quite neutral, when placed as intended, in a room with certain acoustic properties. The obviously better solution seems to be to correct the radiation pattern of the speaker, and end up with a speaker that works well in any room. And it does - to some degree.

Here I show you the frequency response of a system in a typical living room, this is a recent install. Looks reasonably good. But this graph does not say much about the sound of this system. It misses how the room affects the response in the time domain, and how the spectral balance of the decay in this room affects tonality.

The owner of this system lives nearby, so he has heard my rooms many times. He does not need a blind-test to know which is preferred - the treated room, or his larger untreated space. The difference is extreme. But fixing the room is not an option, for practical reasons. So this is my challenge, as a system designer - make it work in this room, as it is. Design a speaker that works in any room, and find methods for placement and calibration.

The speakers are the most important here, and flat on-axis + smooth/controlled pattern works better across a majority of different spaces, that is what I have found. And I have measured and listened to a lot of different rooms.

I chose to implement some smaller minimum-phase frequency corrections in the midrange, and this improved tonality. Room correction with FIR/phase-correcting filters can not fix this.

Then I suggested moving the speakers out into the room, closer to the listening position, and even have an alternative listening position closer to the speakers, for those occasions that the best possible sound is to be experienced. And it works, the sound is then closer to my treated rooms. Why it works is no mystery.

Here is the graph that really tells nothing, perhaps other than that the speakers indeed have smooth, flat frequency response. Still, this is what most people obess about:

 

[ernestcarl]

Was that in the big ceramic tiled room?

The fellow member who shared the measurement (taken some years ago) only stated that it was an acoustically inferior "concrete box" of a room -- he has since moved to a better home with improved room acoustics.

 

[steve59]

I read the first page and jumped to the end. My room my gear I bought a pair of revel F52 based on the science marketing and enjoyed them. One of the best soundstaging speakers I'd heard to date, in fact as much as I liked them in my room I figured the ultima would have to be better so I bought a used pair of salon 1's and surprisingly I thought they sounded the same except the salon didn't soundstage(create images out of sound)at all. Still thinking revel was the answer and dealing in the used market I upgraded from the salon to the salon 2, freaking amazing speakers that were a bit to current hungry for the integrated I bought for the salon 1's. As much as I liked the salon 2 they didn't soundstage in my room either so instead of buying a more powerful amp I sold the salon 2's and went another direction. If the narrow deep waveguide of the f52 was the reason I fell for revel and the changed shallow wide wg is why 1 kind 'staged in my room and the other didn't (the salon 1 is flush mount no wg)is above my paygrade. I buy speakers for the 2 channel image they produce and if that illusion can hide other deficits that's ok. I may be the outlier and nobody can say I didn't give revel a chance. IDK what IDK, but i do know what I want.

 

[ernestcarl]

The idea might be to offset the radiation pattern narrowing in the top octave which is typical of conventional loudspeakers.

I believe it had something to do with improving perceived "spaciousness" or "envelopment" in the context of adding room treatment -- maybe? not totally sure *so I could be totally wrong. I tried to look for the exact video discussion to review the context, but no luck -- he has too many hours-long video discussions posted on YT for me wade through one-by-one right now, sorry.

 

[levimax]

After reading about all the problems with stereo I disconnected one speaker, set DSP to downmix to mono, and slid the other speaker to the center and have been listening to some of my mono LP's. At first it is a little weird but after awhile I am starting to like it. Stereo to me is sometimes like an old fashioned 3-D viewer or 3-D movie where while there is "depth" it seems kind of unnatural and sometime I feel like my brain is straining to process the "unnatural" information. Pure mono, especially recordings made to be listened to in mono, while not "immersive" are very easy to listen to and while it could be "bias after reading the article" the sound also seems more detailed ( I am noticing the LP's pre-echo much more than I usually do).

 

[Duke]

After reading about all the problems with stereo I disconnected one speaker, set DSP to downmix to mono, and slid the other speaker to the center and have been listening to some of my mono LP's. At first it is a little weird but after awhile I am starting to like it. Stereo to me is sometimes like an old fashioned 3-D viewer or 3-D movie where while there is "depth" it seems kind of unnatural and sometime I feel like my brain is straining to process the "unnatural" information. Pure mono, especially recordings made to be listened to in mono, while not "immersive" are very easy to listen to and while it could be "bias after reading the article" the sound also seems more detailed ( I am noticing the LP's pre-echo much more than I usually do).

You might try reconnecting the second speaker and placing it behind or beside the first, but facing backwards. This would essentially be a bipolar system... same first-arrival sound as a single speaker, but now you'll have considerably more energy in the reflection field. If you have a balance control, you can even play around with how loud the rear-firing speaker is in relation to the front-firing speaker. Imo ideally the speakers would be about five feet out from the front wall so that the "backwave" energy doesn't arrive too early. I'd rather not post what I think you might hear because I don't want to risk introducing an expectation bias.

 

[Bjorn]

It took several coffee-breaks to get through this text. And though I disagree with some parts, I find there is much more common ground. To go into detail is beyond this random forum discussion.

So instead I make my comment on something we do agree upon. With a little addition at the end.

Flat on-axis frequency response looks like a no-brainer, and it is not very difficult to achieve. Yet many speakers show quite strange responses, many of which are highly rewarded. Are we missing something here? I think not. Some speakers are tuned to give a reasonably neutral tonal balance in-room, and if the off-axis is colored, the on axis needs to be adjusted. Then you end up with a speaker that can be quite neutral, when placed as intended, in a room with certain acoustic properties. The obviously better solution seems to be to correct the radiation pattern of the speaker, and end up with a speaker that works well in any room. And it does - to some degree.

Here I show you the frequency response of a system in a typical living room, this is a recent install. Looks reasonably good. But this graph does not say much about the sound of this system. It misses how the room affects the response in the time domain, and how the spectral balance of the decay in this room affects tonality.

The owner of this system lives nearby, so he has heard my rooms many times. He does not need a blind-test to know which is preferred - the treated room, or his larger untreated space. The difference is extreme. But fixing the room is not an option, for practical reasons. So this is my challenge, as a system designer - make it work in this room, as it is. Design a speaker that works in any room, and find methods for placement and calibration.

The speakers are the most important here, and flat on-axis + smooth/controlled pattern works better across a majority of different spaces, that is what I have found. And I have measured and listened to a lot of different rooms.

I chose to implement some smaller minimum-phase frequency corrections in the midrange, and this improved tonality. Room correction with FIR/phase-correcting filters can not fix this.

Then I suggested moving the speakers out into the room, closer to the listening position, and even have an alternative listening position closer to the speakers, for those occasions that the best possible sound is to be experienced. And it works, the sound is then closer to my treated rooms. Why it works is no mystery.

Here is the graph that really tells nothing, perhaps other than that the speakers indeed have smooth, flat frequency response. Still, this is what most people obess about:
View attachment 193199

The frequency respons is only the most important factor as long you have a decent response without major deviations or if for example by comparing both in a mediocre/poor acoustical room. This quickly changes when the responses don't deviate too much and you have other aspects between two speakers with very audible differences. Or if you compare it in a highly well treated room vs a poor or mediocre one, and where the time domain matters far more.

Like someone said: "A modest system in a "good" room can easily outperform upscale, expensive equipment (and objective good) in a bad one."

 

[fineMen]

... To go into detail is beyond this random forum discussion. ...
... how the room affects the response in the time domain, and how the spectral balance of the decay in this room affects tonality. ...
... have an alternative listening position closer to the speakers, for those occasions that the best possible sound ...

This thread first discussed, if "the score" can be trusted. You added an interesting aspect. Reflections have to be considered as detrimental for the sound stage realism. The score doesn't take the width of the dispersion pattern into consideration. If I get You right, the score doesn't tell it all, hence shall not be taken as a final verdict.

You reintroduce the argument with mentioning the room affecting time domain, decay and tonal balance--the main concern of the score, conclusively. You then advertise a closer listening position, which naturally would mute the reflections relatively.

This is a side-move, me thinks. I don't want to analyse its persuasive strength. I would rather like to come back to Your sound-stage argument. This could be valid, indeed. The score relies on mono listening tests, and there should be doubt if it was applicable for stereo in the first place!

I lately expressed my expectation, that reverberation in contemporary recordings is either recorded separately, or is generated synthetically, and in either case added to close-proximity, individual recordings of the solo instruments. Other, additional techniques might apply.

Now, if the reverberation isn't stereo in itself? The reverberation, as said recorded separately or synthesised, is basically mono, added without further ado to both stereo channels without differentiation. What then about the time relation of in-room reflections to the direct sound? And by which means would these affect the plausibility of the stereo imaging?

I personally argue, that reverberation, by our all hearing, isn't evaluated for directional cues. Experiment: close an ear and try, blindfolded, to orient Yourself confined in a room. Possible - proven! The notorious two-ear technology isn't needed to get the picture. Maybe it is just the significant burst of the early and late arrivals, a timely, but not directional pattern of reflections, that portrays the room.