Questions about speaker measurements and what they represent or don't

[mglobe]

Thanks. IOW mglobe seems not to have properly predicted "what you would expect from FR" .


(In a two-way comparison a difference plot for each measurement might be even simpler to interpret, though maybe harder to generate)

Ok - but how well do the measurements comport with reality for you?

Personally, I certainly have found narrower dispersion speakers to work in more rooms than wider dipersion ones, but then again… a lot of rooms sorta sucked.

The measurements are consistent with reality to my ears in terms of FR. The magnitude of the overall improvement in sound reproduction is greater than what the measurements suggest to me. The Reference 1 Meta's are a LOT better sounding than the R3's (duh I guess).

Please don't believe and/or suggest from these comments that I don't buy into measurements. That couldn't be further from the truth.

 

[kemmler3D]

1) transient response - how can I tell from the measurements that a speaker is capable of producing transients properly?

Look for flat and extended frequency response, (high and low) short and coherent step response (basically the same thing as transient response), lack of resonances, and if group delay measurements are available, a smooth and low group delay across the board.

2) depth of field - quested v2108. this was the speaker that started it all for me.

"depth of field" is pretty subjective to relate to measurements, but I would say your best bet will have to do with smooth dispersion / directivity.

what measurement represents the speaker ability to portray a vast depth of field across the frequency spectrum?
3) erin's horseshoe stage - I am not sure this exist. this might be more tied to speaker positioning.
further apart and more energy is perceived from the side compared to a narrower speaker positioning that you perceive as more level coming from the middle of the picture.
there was a video of brian lucey (exceptional mastering engineer) that kinda says the same thing.

This depends on the room but you're going to want smooth directivity and probably wide across the frequency spectrum also, but not sure.

4) overall resolution - what measurements represents the amount of detail you can display (see what I did there?) while reproducing sound? some speakers I can't see the tree within the forest

Low distortion, extended flat frequency response, sometimes you will also associate this with elevated treble.

Welcome to ASR!

 

[krabapple]

The measurements are consistent with reality to my ears in terms of FR. The magnitude of the overall improvement in sound reproduction is greater than what the measurements suggest to me. The Reference 1 Meta's are a LOT better sounding than the R3's (duh I guess).

Please don't believe and/or suggest from these comments that I don't buy into measurements. That couldn't be further from the truth.

Bass performance plays an outsize role in user perception of loudspeaker sound quality. We know this from the work of OLive, Toole et al., which IIRC indicates that it accounts for as much as 30% of 'preference'.

I did not suggest you don't 'buy into measurements'. I suggested that your idea of 'what to expect' from them , in terms of predicting sound quality, was incorrect, based on the comparative graphs AnalogSteph posted. Klippel measurements of these two clearly shows one putting out more bass than the other,as well as several dBs worth of difference in critical treble bands. So one might expect the difference to be significant -- perhaps one even sounding a LOT better. He also noted that their use cases are different, with R3s being intended to exploit room boundary effects. So comparing them properly would require taking that into account too, and might make the difference fade in significance.

 

[phoenixdogfan]

View attachment 399491
Let me guess: The R1 Metas have quite a bit more bass and less mid-treble brightness? People tend to underestimate the (literal) impact of bass. They also have a slight midrange dip around 1-2 kHz, so the R1 Meta which arguably is a hair too far in the other direction might have a bit of midrange "honkiness" in comparison.

The R3s are tuned for placement close to a back wall, while the R1 Metas can be used freestanding without any issues. An adjustment of room EQ will be required when swapping from one to the other.

The bass response is the most obvious one. The f3 on the R3 Meta's is 77hz while it's 45hz on the Reference 1 Metas. That is almost an octave difference, and will be immediately audible. The dispersion is also a major difference. The 6db zone for the R3's is +/- 40 degrees while is +/- 50 degrees for the Reference 1's. Finally, at volume the Reference 1's hold up all the way to 50 hz at 96 db while the R3's start to show compromised dynamic range as high as 150 hz.

As a result, the Reference 1's will sound much more like a full range floor stander that can be played quite a bit louder than the R3's, and it will have a more panoramic sound stage because of its wider dispersion. The Reference 1 Meta will also sound a little more laid back than the R3 because of a trough of 1-2 db from 2-4 khz, though that can eq'd to the user's taste on either speaker because both speakers have very smooth directivity and will be readily adjustable in that area.

So, in spite of the score, the Reference 1's will sound like the better speakers because they are.

 

[mglobe]

The bass response is the most obvious one. The f3 on the R3 Meta's is 77hz while it's 45hz on the Reference 1 Metas. That is almost an octave difference, and will be immediately audible. The dispersion is also a major difference. The 6db zone for the R3's is +/- 40 degrees while is +/- 50 degrees for the Reference 1's. Finally, at volume the Reference 1's hold up all the way to 50 hz at 96 db while the R3's start to show compromised dynamic range as high as 150 hz.

As a result, the Reference 1's will sound much more like a full range floor stander that can be played quite a bit louder than the R3's, and it will have a more panoramic sound stage because of its wider dispersion. The Reference 1 Meta will also sound a little more laid back than the R3 because of a trough of 1-2 db from 2-4 khz, though that can eq'd to the user's taste on either speaker because both speakers have very smooth directivity and will be readily adjustable in that area.

So, in spite of the score, the Reference 1's will sound like the better speakers because they are.

Thanks for that. A lot of what you describe is exactly what I hear. The reason I’ve dismissed the bass performance difference some is because I’m crossing the Reference speakers at 60, while I was crossing the R3’s at 80. I moved my subs and seating position a bit to take care of a 100hz dip, but even before doing that the Reference were quite an improvement. No question the lack of compression in the Reference plays a part, and yes they play a lot cleaner at a lot higher SPL. It’s not subtle. I suspect there’s also less resonance in the Reference, and likely less IMD as well.

 

[sam_adams]

Nice flypaper discussion . . .

 

[Blumlein 88]

Is anyone forgetting you can know impulse response from frequency response and vice versa?

 

[Keith_W]

Welcome to ASR. I am going to refrain from attacking you because I believe you are asking these questions in good faith.

To me, your questions reflect a desire to understand and correlate subjective perception to objective measurements. This is a HUGE topic that can not be covered in a single thread. For example, what correlates with "transient attack" or "dynamics" encompasses a few measurements, including:

- phase distortion. A speaker that distorts phase will deconstruct a transient and smear it over time, reducing both its attack and prolonging the decay.
- group delay (related to phase distortion). If one driver is delayed with respect to another, the attack of the transient will seem smeared.
- speaker compression and volume nonlinearity. If you command 75dB, you should get 75dB. If you command 85dB, you should get 85dB. Unfortunately speakers don't necessarily do this, sometimes you get 82dB. If speakers compress, it will not reproduce the transient properly.
- room acoustics. If reflections are loud and early enough, it will smear the sound. If room reverberation is bad enough, it will bleed into the next transient, creating a muddy sound.

I could go on. But you get the idea.

The problem with measurements is that it is not intuitive what they represent. As I have shown, several different measurements may correlate with one subjective phenomenon, and that's with a phenomenon where the meaning is known! A lot of study has to go into interpreting measurements. And even then it is not enough, you need an understanding of psychoacoustics to know what measurement is correlated with which subjective phenomenon.

Another problem with measurements is that they may not be representative. Microphones and measurement techniques themselves have failings. Toole discusses this extensively in his book. For example, omnidirectional microphones are unable to hear direction, whereas humans were evolved to hear direction.

I recommend that you start off by understanding what all the measurements mean. Very slowly you will eventually be able to correlate measurements with what you hear. Buy yourself a copy of Toole.

 

[Holmz]

The measurements are consistent with reality to my ears in terms of FR. The magnitude of the overall improvement in sound reproduction is greater than what the measurements suggest to me. The Reference 1 Meta's are a LOT better sounding than the R3's (duh I guess).

Yeah and Toole says that FR is the most important.

Please don't believe and/or suggest from these comments that I don't buy into measurements. That couldn't be further from the truth.

I can read, and, in fact… you mentioned that earlier, in a very clear fashion.


What I am saying that is that generally most people here do not give a damn about phase and time domain behavior.
If you have two speakers with identical behavior except that one is different in the time domain, then they should sound different.

There seems to be two, or maybe three, choices:

• Get a speaker that is flat in the Frequncy domain, and fix the time-domain with DIRAC (FIR EQ)
• Get a speaker that is good in the time domain and fix the frequency domain with EQ, maybe IIR/PEQ EQ.
• Get a speaker that is not good in either and fix the response with a FIR EQ

Speakers with higher order cross overs, and higher order box designs, generally have worse transcient response, and DIRAC cannot remove that.

 

[williamr]

Is anyone forgetting you can know impulse response from frequency response and vice versa?

Can we assume linearity within the regime of the measurements for all speakers?

I'm asking as a neophyte. I realize at the engineering level, speaker components and the electronic components are typically used in the linear regime and that DSP often is used to help linearize the system, but wouldn't an impulse response side-by-side with the FR help assure this?

 

[Blumlein 88]

Can we assume linearity within the regime of the measurements for all speakers?

I'm asking as a neophyte. I realize at the engineering level, speaker components and the electronic components are typically used in the linear regime and that DSP often is used to help linearize the system, but wouldn't an impulse response side-by-side with the FR help assure this?

They are usually linear. RayDunzl posted here the REW impulse response from a sweep and then he sent a 10 hz square wave as a step response. The results were the same. It is a test you could do to confirm.

Importance of impulse response

Pardon the interruption, but... Long ago it bothered me that software could take a sine sweep (no edges) and create a measurement (impulse or step) that does have "edges". So I set out to find the difference between "real" and "calculated" values. Real: Single Full Scale Bit sent through...

www.audiosciencereview.com

 

[Holmz]

Is anyone forgetting you can know impulse response from frequency response and vice versa?

Yeah no.
In theory yeah, but it is only the amplitude that is ever shown.

One really needs the phase, or imaginary part to get back to the time domain. You cannot do it with just amplitude alone.

 

[cavedriver]

You have to try to be constructive in life and not insult the members of a forum to start with.
Genelec makes speakers as they should be done, from recognized loudspeakers like the PHL (also unfortunately used by Stenheim). Their prices are out of reach for me and I don't think my hearing is worth this expense. And I'm more into DIY.
Blindly, I can like a whole bunch of cheap products even very amateur ones.
Why? Because even very low-end devices like amps, CD players or streamers will transmit 90% or more of the musical message quite easily. Which does not mean that in terms of measurements, more expensive devices are better. But there remains the threshold of audibility.

Just found this post after listening to some Stenheims and wondering what else you can tell me about how Stenheim uses the PHL drivers or who else uses them? I'm intrigued by their woofers because of their fairly high efficiencies, but was told by a Stenheim dealer that S does their own doping or specs custom doping on the cones. Having now looked closely at the cones I suspect S is simply using off the shelf PHL drivers. I am glad to see that there is a retailer in France where the general public can buy PHL drivers but I haven't yet found any published DIY designs or other manufacturers that make high efficiency designs that don't cost an arm and a leg.

 

[rank]

I only know Pascal Louvet but it's expensive too (their 5 Element V2 Esprit remains an accomplishment: 4 cabinets, 240 kg -!-, 400W, 97 dB/W/m, €38,900 alas). Palladium Audio too but it's custom sound-systems so even more expensive.
The PHL cones are already coated against humidity and UV, I haven't studied the load or filtering at Stenheim so I can't tell you more. Otherwise to go further, they could ask like Martin Audio for special productions. But again, the cost...