Relative importance of different speaker metrics?

[garbz]

We are bombarded with a lot of specs and measurements here with these amazing speaker reviews, but how do you objectively compare different speakers?

I'm thinking here something like the Dutch and Dutch 8Cs which Erin reviewed, and their performance in terms of dispersion is simply astonishing, and the cardioid design doubly so, and frequency response triply so. But distortion... simply insanely high compared to even my B&Ws which run at 1/4 of the price.
But my B&Ws have a very bright frequency response and crappy dispersion characteristics. The cabinet is very devoid of resonances though.
Now go to another speaker with good dispersion, reasonable frequency response but crappy cabinet resonances.

Is there any guide as to what matters? A pecking order of how to understand all the metrics to chose an objectively good speaker?

I should add that we're talking about a standard house listening environment here, I don't want to get into compensating dispersion characteristics with room treatment etc. And obviously the power handling and volume is an easy one for people who love loud.

 

[coonmanx]

Yup. Good frequency response and dispersion are at the top of the list.

 

[Digby]

Is there any guide as to what matters? A pecking order of how to understand all the metrics to chose an objectively good speaker?

I should add that we're talking about a standard house listening environment here, I don't want to get into compensating dispersion characteristics with room treatment etc. And obviously the power handling and volume is an easy one for people who love loud.

This is the problem, isn't it? What matters the most - does x matter more than y and, if so, is there a situation where y could matter more than x (bad performance on z attribute, for example). Maybe x matters more to you personally (subjective), but y to me.

Overall, I do believe there are some factors that are likely to matter more to the typical listener and I think Earl Geddes has done some research into this.

I have had recent experience myself with this conundrum. A pair of (admittedly rather smaller) speakers that measured very well here, did not perform as well to my tastes, as a pair of rather larger speakers that are around 1/4 of the price. Having heard the larger speakers the deficiencies of the smaller ones, however much they may be better in some respects, did not outweigh the negatives, which once heard I couldn't unhear.

With regards to power handling and volume, I don't think it is just a question of loud, but of how well controlled a speaker is for a given volume. You probably want a good amount of overhead to account for crescendos on top of whatever you would consider a loud volume, all of this while sounding under control.

 

[garbz]

I'm not even so much talking about personal tastes as much as what objectively screws up the signal as it gets to us. Consider the following:
Headphones: Diaphragms right close to the ear meaning no interaction with much of the outside world. THD here translates to THD on your eardrum. Frequency response would be nice and flat, but more important would be a lack of resonance allowing headphone to be EQed.

But speakers... sound that comes to your ears bounces all over the place first. So if you have a low THD speaker with crap directivity you end up with garbage at your ears, but what about a high THD speaker with really good directivity?

Good frequency response and dispersion are at the top of the list.

Can you expand on that? Is the problem the room interaction is worse than distortion issues in the speaker itself?

 

[coonmanx]

Good frequency response is just that. You won't be missing anything in any part of the sound spectrum.

Dispersion means that if you are off axis you still get a very good sound and possibly even good imaging. I rarely sit in the "sweet spot" anymore.

Also where I listen to the speakers, the sound is not "bouncing all around" since there is nothing for it to bounce off. And if that is a problem then the individual listener can always use room treatments to correct for that.

 

[coonmanx]

I don't worry at all about speaker distortion. To me, it means nothing at all since all of my speakers have quality drivers in them. I don't listen to crap speakers.

 

[Koeitje]

Good frequency response is just that. You won't be missing anything in any part of the sound spectrum.

Dispersion means that if you are off axis you still get a very good sound and possibly even good imaging. I rarely sit in the "sweet spot" anymore.

Dispersion is important because you don't only listen to the sound straight from the speaker but also to reflections.

 

[coonmanx]

Dispersion is important because you don't only listen to the sound straight from the speaker but also to reflections.

It isn't so much reflections as the ability of a speaker to sound good off axis. If reflections are an issue then room treatments are available to deal with them. But if you have a carpeted floor you probably won't be getting a lot of reflection from that. Reflections are usually seen as a negative.

 

[coonmanx]

Here is a weird thought about reflections...

I recently gifted a friend an integrated amplifier and a pair of speakers that I built. I'm not sure why I built those speakers because I already had too many pairs of speakers. So I guess it was just a fun project where I could learn something.

Anyway, they are sitting on a wooden floor so there are probably all kinds of reflections going on. Yet those speakers sound pretty damn good in that particular room. I don't know why but every time I have been over there listening to music it always sounds very good. Not raised off of the floor at all. Not ideal at all...

 

[garbz]

Room treatments are out of scope. As I mentioned standard house environment. Let's assume you have a living room and not a dedicated listening space making room treatments out of the question.

I challenge your "reflections are seen as a negative" comment. They contribute massively to a widened sound stage and it is most definitely possible to over treat a room and make it sound "dead".

Also I don't get your comment. THD means nothing to you because you don't have crap drivers. THD may or may not have anything to do with the "crapness" of your drivers. THD can come from a desire to listen loud (all speakers will distort at some point), it can come from the design of the speaker, the cabinet, there are speakers above $10k which have significantly higher distortion than speakers worth $1k and that is not because they chose "crap drivers" and definitely because they aren't "crap speakers".

 

[sgent]

Watch Erin's interview with Geddes. They get more in depth, but discuss distortion and its meaning quite a bit. All of Gedde's papers are on his website and may be worth reading as well. The gist of what I got from it is that distortion almost doesn't matter when compared to the importance of directivity and frequency response.

 

[dfuller]

In order:
1. Frequency Response on axis (includes flatness, smoothness i.e. lack of resonances, LF extension)
2. Frequency Response horizontally off axis (horizontal directivity)
2.5: Frequency Response vertically off axis (vertical directivity) - this is far, far less important than horizontal
3. Nonlinear distortion

Speaker distortion is generally almost always low order (2nd and 3rd, primarily) and thereby almost totally masked. Good speakers keep it low enough that it is completely inaudible in program material (generally on the order of 0.1% at 86dB above ~80hz).

 

[Sancus]

The D&D 8C indeed shows high distortion, but it's 2nd and 3rd harmonic(harder to hear) with the 4th and 5th 15-20dB lower. Erin did not report hearing any audible distortion with SPLs as high as 105dB@4m. He also noted that there are no mechanical issues/noises which he considers more typically audible.

We don't have a very clear picture about how masking due to duration, other tones, and distortion interact, unfortunately. You can find information easily enough that's been tested with pure tones, but those are not very useful because there are a lot of masking effects, both frequency and time-based.

There's a (relatively, 2019) recent paper on this topic that looks potentially very interesting but I dunno if I want to pay for it lol.

 

[thewas]

In order:
1. Frequency Response on axis (includes flatness, smoothness i.e. lack of resonances, LF extension)
2. Frequency Response horizontally off axis (horizontal directivity)
2.5: Frequency Response vertically off axis (vertical directivity) - this is far, far less important than horizontal
3. Nonlinear distortion

For people that use EQ I would even put 2 and 3 before the flatness part of 1.

Speaker distortion is generally almost always low order (2nd and 3rd, primarily) and thereby almost totally masked. Good speakers keep it low enough that it is completely inaudible in program material (generally on the order of 0.1% at 86dB above ~80hz).

IMD can get in my experience quite audible though if the rest is nice and smooth like for example in my LS50 pairs when SPL gets too high for them.

 

[dfuller]

For people that use EQ I would even put 2 and 3 before the flatness part of 1.


IMD can get in my experience quite audible though if the rest is nice and smooth like for example in my LS50 pairs when SPL gets too high for them.

Yeah though at the point that IMD starts being audible with the LS50s the speaker is also producing a ton of HD anyway.

 

[thewas]

Yeah though at the point that IMD starts being audible with the LS50s the speaker is also producing a ton of HD anyway.

The high HD though of the LS50 is the bass region while the audible IMD is higher.

 

[dfuller]

The high HD though of the LS50 is the bass region while the audible IMD is higher.

Huh, interesting. Wonder if it has to do with it being coax and the midwoofer doing relatively high excursions?

 

[thewas]

Huh, interesting. Wonder if it has to do with it being coax and the midwoofer doing relatively high excursions?

Possibly too but I think most audible are the imd of the midwoofer in the mids like on most 2-ways.

 

[garbz]

Thanks for the links and the video. Have some good reading material for this weekend