Which anechoic response curve predicts the preferred loudspeaker? (fun exercise)

[preload]

Here are the FR curves for 3 different speakers, as published in Stereophile. It's the usual "anechoic response on tweeter axis at 50", averaged across 30 deg horizontal window and corrected for mic response, with complex sum of nearfield woofer and port responses plotted below 300Hz). Pretend you'll add a sub, so ignore the bottom octave.

Which would you predict would sound the best, 2nd best, 3rd best? How confident would you be based on your analysis? (P.S. If you recognize the speaker identity based on any of these curves, please don't respond, it wouldn't be fair).

Edit: added anechoic response windows

Speaker A - anechoic FR averaged over 30 deg horizontal window

Speaker A - horizontal response family, normalized to tweeter axis, -90 degs to +90 degs

Speaker B - anechoic FR averaged over 30 deg horizontal window

Speaker B - horizontal response family, normalized to tweeter axis, -90 degs to +90 degs

Speaker C - anechoic FR averaged over 30 deg horizontal window

Speaker C - horizontal response family, normalized to tweeter axis, -90 degs to +90 degs

 

[aarons915]

There really isn't enough info in the stereophile listening window to reliably answer that but only based on these I would rank them in the order you posted them, A, B then C.

 

[pozz]

Nonanechoic measurements, no full spin data. Can't form an opinion.

 

[preload]

Nonanechoic measurements, no full spin data. Can't form an opinion.

These were anechoic.

 

[pozz]

Point is that accuracy is limited. These aren't done in a controlled chamber and there's no off-axis information.

This is why people say "flat FR is meaningless", the pretense being that the data above is comprehensive. Those single curves say very little about the overall radiation pattern.

 

[MZKM]

It’s just a horizontal listening window response, so not enough data, but from just those images I guess A.

 

[preload]

Point is that accuracy is limited. These aren't done in a controlled chamber and there's no off-axis information.

This is why people say "flat FR is meaningless", the pretense being that the data above is comprehensive. Those single curves say very little about the overall radiation pattern.

It’s just a horizontal listening window response, so not enough data, but from just those images I guess A.

Okay, fair enough. I added in the (horizontal) off-axis response curves as well from Stereophile. Many of the spinorama curves seem to be derived from these curves, if that helps.
My understanding is that these are quasi-anechoic, and JA uses a MLS technique based on a windowed impulse response measurement. Close enough.

I also believe that when developing his preference formula, Olive had all of the usual spinorama data available. However, what ended up in his equation was the predicted in-room response (a weighted average of the on-axis, early reflected, and sound power measurements) and the on-axis response, separately. So I'm thinking that if it's possible to make predictions of loudspeaker preferences based on measurements, I'm providing a whole lot of measurements!

BTW, this is just for fun. If I get enough responses, I'll share the identity of the speakers and my long-term listening impressions of two of them.

 

[Sancus]

Speaker B, A, then C to me.

Speaker B looks to have almost perfectly even off-axis response up to 10khz and is also extremely wide dispersion.

Speaker A looks decent, but isn't as wide nor even in the off-axis. It seems OK though.

Speaker C just looks pretty poor all around.

Adding: I agree with the responses that would indicate you couldn't tell anything from the horizontal window responses alone btw, to me the off-axis graphs are by far the key in interpreting measurements.

 

[Dennis Murphy]

I'm pretty sure I know which is which, so I won't comment other than to note that If B turns out to be a dog, it wouldn't speak very well for the Spinorama protocol.

 

[preload]

Very interesting! And for those responding, if you have the time and are willing, could you also share what you're seeing in the curves that makes you think one speaker will sound better than the other?

 

[dasdoing]

personaly it makes no sense to see "averaged across 30 deg horizontal window" only

 

[MZKM]

Okay, fair enough. I added in the (horizontal) off-axis response curves as well from Stereophile.

I’ll switch my ranking from ABC to BAC.

 

[Absolute]

BAC.

The B would likely be preferred because the off-axis response is incredibly smooth, also very wide.

The A looks decent off-axis, but have far too much energy (broad dispersion) from 3-8 khz off-axis that will likely result in a very top-heavy sound in-room.

C looks like a full-tone speaker or something with a huge driver being forced to play way too high in frequency because it starts beaming like crazy at around 1500 hz.

Based on on-axis alone I'd pick ABC.

 

[Plcamp]

Constraining to +/-30 degrees, “A” looks better than “B” due to rolloff off axis above 10 kHz. But if you are as old as I am, it might not matter?

 

[Sancus]

Constraining to +/-30 degrees, “A” looks better than “B” due to rolloff off axis above 10 kHz. But if you are as old as I am, it might not matter?

Even if you can hear it, there is so little energy in music above 10khz that its relevance is fairly minor, IMO. It doesn't contribute anything to spatial qualities. The peaking in "A" from 3 to 7khz, a very audible region, is a bigger sin than anything happening above 10khz.

Which is not to say that "A" is terrible, but it has that issue and has narrower dispersion which is also very audible.

 

[preload]

personaly it makes no sense to see "averaged across 30 deg horizontal window" only

Would it help to see the averages over the vertical window? Didn't want to overcrowd.

 

[preload]

Speaker C just looks pretty poor all around.

How would you predict Speaker C to sound based on your interpretation that the measurements look "pretty poor all around?"

 

[amirm]

My understanding is that these are quasi-anechoic, and JA uses a MLS technique based on a windowed impulse response measurement. Close enough.

Windowed response only gives you the mid to high frequency response. For bass JA makes near field measurements of the woofer which exaggerates the woofer response in low bass that varies with configuration/size of the speaker:

That is where the hump in 100 to 200 Hz is coming from.

So no, they are not good enough to determine tonality.

You need proper early-window energy to predict in-room response which is also missing.

His measurements are good to find horrible speakers but distinguishing between similar ones is much more difficult.

 

[Sancus]

How would you predict Speaker C to sound based on your interpretation that the measurements look "pretty poor all around?"

Not totally sure tbh, the off-axis peaking/on-axis dip around 5khz could be annoying. The whole region between 3-10khz looks like it has too much energy. So it may be bright, though seeing a PIR would confirm this. It has very narrow dispersion too, so it probably sounds small in room.

Calling it pretty poor may be overly critical. I don't know what speakers the others are, but they look like they are very good ones. C's off-axis response is not a total mess, and it may have intentionally narrow dispersion for use cases that prefer that(ie: studio monitors). I don't know.

E: Edited the first paragraph a bit.

 

[dasdoing]

Would it help to see the averages over the vertical window? Didn't want to overcrowd.

i am in the "only sweet spot sounds good" team, so on axis for me is above all