Mono vs Stereo Speaker Evaluation

[BenB]

First a little background:
I've been building speakers with limited vertical dispersion, but typical (to wide) horizontal dispersion. I accomplish this by using nested line arrays and careful crossover optimization for both on-axis and off-axis response. This typically makes the speakers very tall, but I managed to fit my latest design in a 34" tower cabinet (though I still have to put them on stands to get the ribbon tweeter at ear height... so it's more of a tall bookshelf the way I use it). One of the reasons I chose to make this (somewhat compromised) design was because it would allow for listening to the speaker as intended (vertically), and also horizontally.

By alternately listening to the speaker vertically and on its side, the listener gains an appreciation for the impact of the stronger reflections: When the speaker is vertical, the lateral (sidewall) reflections are much stronger than the floor and ceiling reflections. When the speaker is horizontal, the floor and ceiling reflections are much stronger than the sidewall reflections. What's great about this is that it isolates so many variables because it's the same speaker(s) both times: same drivers, same crossovers, same limits of distortion, same bass extension, etc.

I first performed this comparison (vertical vs horizontal) in stereo, and I found a pretty dramatic change in the soundstage. With stronger sidewall reflections, the soundstage was much wider. My subjective reaction was that the music was more engaging and more authentic this way: less like listening to a pair of speakers and closer to an actual performance. I expected to be able to perceive more detail with the speakers in the vertical position, but I didn't notice a substantial change in this regard. Although I had a strong preference for wide horizontal dispersion, I could understand why someone else might prefer a limited horizontal dispersion, which sounded a little simpler, cleaner, drier: like getting just what's in the recording itself.

I later performed the same test (vertical vs horizontal listening) with a single speaker. The differences were much, much less pronounced. The differences were now so subtle, that I didn't bother trying to identify a preference.

Conclusion:
Mono (single-speaker) listening didn't provide the opportunity to assess (or appreciate) differences that caused dramatic changes in the perception of soundstage when listening in stereo. For me, this casts serious doubt on the claims that single-speaker evaluations are sufficient.

Considerations:
The test was sighted. Only 2 people have performed this exact test. (I believe the results were similar, but the 2nd person can choose whether or not to chime in here since he reads this forum.) Most speakers don't have such drastically different dispersion characteristics (vertical vs horizontal), and even "narrow" dispersion speakers aren't typically as narrow as these speakers are vertically (though that depends a bit on whether you concentrate on the large off-axis angles or the small off-axis angles).

I have attached some reference pictures to show the speakers used for the test, as well as frequency response plots to compare vertical and horizontal dispersion. I have purposely added a small downward tilt to the upper mids and treble that I find to give a more pleasing balance. In my experience, spectrally flat speakers with limited dispersion are often perceived as bright and can cause listening fatigue. Of course, most on this forum would EQ to taste, but I've been lending these speakers out to enthusiasts and non enthusiasts, and I wanted a balance that would be inoffensive even with less-than-perfect recordings. The frequency response plots show on-axis in blue, with horizontal dispersion in red, and vertical dispersion in green. I have included plots for 45 and 90 degree off-axis measurements.

 

[DonH56]

I've had Toole's book since the first edition and other references (AES etc.) since well before. If you've made up your mind and have not read the background of the mono test methodology and resulting mass of test data there's not much point in discussing any further.

Since Dr. Toole is an ASR member perhaps he'll chime in but all too often it is tilting at windmills.

 

[Dzhaughn]

I don't see how you can say that the vertical oriented speaker is in the same position in the room as the horizontal oriented speaker. For one thing, your room is not likely symmetric in the vertical/horizontal axis. For another, your speakers are very asymetric vertically and horizontally.

So you have shown that a different position in a room may affect measured sound and listener preference. Good. But we knew that.

 

[aarons915]

By alternately listening to the speaker vertically and on its side, the listener gains an appreciation for the impact of the stronger reflections: When the speaker is vertical, the lateral (sidewall) reflections are much stronger than the floor and ceiling reflections. When the speaker is horizontal, the floor and ceiling reflections are much stronger than the sidewall reflections. What's great about this is that it isolates so many variables because it's the same speaker(s) both times: same drivers, same crossovers, same limits of distortion, same bass extension, etc.

Curious about your impressions when the vertical reflections are stronger. It's no surprise you prefer stronger horizontal reflections but would you say the vertical reflections are beneficial as well? Based on your testing do you think you'd prefer strong horizontal AND vertical reflections or would you prefer only strong horizontal and attenuated vertical reflections?

 

[mhardy6647]

For home theater front channels, one of these is needed...

 

[BenB]

Curious about your impressions when the vertical reflections are stronger. It's no surprise you prefer stronger horizontal reflections but would you say the vertical reflections are beneficial as well? Based on your testing do you think you'd prefer strong horizontal AND vertical reflections or would you prefer only strong horizontal and attenuated vertical reflections?

I haven't been able to run tests that are as controlled as I'd like to answer this question. I have considered using active crossovers on my nested arrays, and switching the top half on and off. When both halves are on, the vertical dispersion is reduced much more than if only the bottom half is on. Of course, I'd have to adjust the tweeter level as well.

Judging what I've heard from different speakers, I think vertical reflections can be pleasing for single-speaker mono, when we are starved for ambiance. In stereo, I think the spectral coloring effect of the vertical reflections may be nice, as it can prevent things from sounding too bright. If you are willing to EQ a speaker with narrow vertical dispersion to your liking, I think you can get the best result. In my experience, vertical reflections help listeners localize sounds to their speakers, making it more difficult for the speakers to "disappear" (where the music sounds completely disassociated from the speakers). My co-worker borrowed the speakers pictured in this thread, and his wife (who had no information about what to expect) opined that the speakers disappeared, and he declared that it was "like being there". He listens mainly to classical and baroque.

I built my nested line array speakers sort of on a hunch, and with 7.75 foot ceilings in my basement (where my theater is), the importance of reducing vertical reflections may be exaggerated. The result is the best speaker I've ever heard, and an experience that's significantly different from other designs. Interestingly, when I've demonstrated my speakers to people, and we stop listening to discuss what we've heard, it becomes apparent that the room doesn't actually have good acoustics. However, the speakers don't energize the offensive vertical reflections, and manage to sound "perfect" despite the low, untreated ceiling (using the words of another audio enthusiast there, not my own.)

 

[BenB]

I don't see how you can say that the vertical oriented speaker is in the same position in the room as the horizontal oriented speaker. For one thing, your room is not likely symmetric in the vertical/horizontal axis. For another, your speakers are very asymetric vertically and horizontally.

So you have shown that a different position in a room may affect measured sound and listener preference. Good. But we knew that.

In order to understand what's different about this test, perhaps you should think of the vertical and horizontal oriented speaker set up as using different speakers. We could call them the SLH and the SLV. The SLH and SLV just happen to use the same drivers and the same crossovers, but they are different. Harman research tells us (or more specifically, the interpretation of the research conveyed by the researchers) that mono testing and stereo testing gives the same preference results, but with stereo testing being less revealing of differences and resulting in more closely clustered scores. Here we have a set of speakers (SLH and SLV) that are perceived as being extremely similar in mono, but with drastic differences in stereo.

 

[patate91]

In order to understand what's different about this test, perhaps you should think of the vertical and horizontal oriented speaker set up as using different speakers. We could call them the SLH and the SLV. The SLH and SLV just happen to use the same drivers and the same crossovers, but they are different. Harman research tells us (or more specifically, the interpretation of the research conveyed by the researchers) that mono testing and stereo testing gives the same preference results, but with stereo testing being less revealing of differences and resulting in more closely clustered scores. Here we have a set of speakers (SLH and SLV) that are perceived as being extremely similar in mono, but with drastic differences in stereo.

What's missing from some Harman plots (I didn't see a lot of them) is what the ranking for each participant (the individual result). Looking at black and white plots there's no indications if a participant got the same result as you. People at Harman certainly have this informations.