Do loudspeakers need to image precisely?

[SIY]

Hi

The issue is complicated. Speakers have a very hard job to do: Reproducing as faithfully the electrical signical that is fed to them.

Where?

See, that's the problem- the signal fed to them is one-dimensional. The soundfield created (and I choose that word deliberately) is 3 dimensional. So by definition, the speaker cannot and will not "reproduce faithfully the electrical signal fed to them."

 

[Juhazi]

"Imaging" is part of the stereophonic (or multichannel) illusion we are trying to achieve. It is more dependant on the room and speaker positioning than the loudspeaker used. Another problem is, that we don't know how a recording sounded in the mastering room, to the producer and mixing engineer.

We listeners are very different in what we expect of "imaging" in stereo. I know many who always play that certain press of Pink Floyd's DSOTM or some Yello stuff to hear how a system performs/images. I think that is funny.

 

[sergeauckland]

"Imaging" is part of the stereophonic (or multichannel) illusion we are trying to achieve. It is more a dependant on the room and speaker positioning than the loudspeaker used. Another problem is, that we don't know how a recording sounded in the mastering room, to the producer and mixing engineer.

We listeners are very different in what we expect of "imaging" in stereo. I know many who always play that certain press of Pink Floyd'd DSOTM or some Yello stuff to hear how a system performs/images. I think that is funny.

Yes and no. I agree that the room and loudspeaker positioning makes a large, possibly the biggest difference, but even in a benign room, what matters is how closely matched the two loudspeakers are. The frequency response of the loudspeaker will set the 'speaker's 'tone', but what matters for imaging is how closely the two loudspeakers are matched. Matching within +-1dB from a few hundred Hz upwards, ideally closer, gives sharp imaging. Sadly too many loudspeakers these days have discrepancies of several dBs even if the overall efficiency is close, and those can't possibly image sharply.

I noticed an improvement in image sharpness when I equalised my loudspeakers to +-1dB and reduced differences to +-0.5dB over what I had before, but I accept that it's not something I could test blind. However, thinking through how phantom images are created, it's clear that if the balance between the loudspeakers varies to any extent with frequency, the phantom image will wander or expand in width compared with two identical loudspeakers.

One of the main benefits, I think, of active DSP crossovers is that the individual drivers can be equalised such that a pair is closely matched.

S.

 

[mitchco]

Re: One of the main benefits, I think, of active DSP crossovers is that the individual drivers can be equalised such that a pair is closely matched.

@sergeauckland agreed. I would add... closely matched in the time domain as well.

 

[Juhazi]

^Obviously you mean "matched to listening spot" This includes reflections of which first order sidewall reflections are dominant in regard to response and imaging. Backwall behing the listener is surprisingly important too, it should be absorptive and diffusive. These reflections are also very specific to location, multipoint eq is always a compromise.

In time domain a reflection is a challenge, how do you handle that with eq/rephase/dirac? How does it measure to the next seat then?

 

[tmtomh]

Very interesting discussion. I think it highlights the limits and flaws of the idea that a home stereo should be judged on the metric of live music performance.

With the exception of chamber music, solo acoustic performance, and perhaps small-combo jazz, recorded music - even including live albums - is not recorded, mixed, or mastered as a live performance.

Conversely, as has been noted above, most live music venues present a listening experience that is technically inferior to home listening with recorded music. The emotional effect of live music might exceed that of recorded music. But a live venue is going to have virtually no "sweet spot"; and most indoor live venues for amplified popular music are too loud, highly distorted, with poor frequency balance and dreadful acoustics.

The goal of good speaker imaging iMHO is about reproducing a recording as accurately as possible. Modern multi-track studio productions almost never are mixed with a soundstage that could be reproduced live - and even when the soundstage mix is physically plausible, the different instruments have been recorded at different times, with their own separate mic setups and room ambience - they simply don't sound the same sonically as a simultaneous live performance with the same placements would.

But such synthetic productions are mixed in stereo, and so it helps if the speakers can reproduce that imaging accurately - less because it's enjoyable to hear instruments or sounds come from some extreme L or R position, and more because good imaging makes it possible to distinguish the various instruments/sounds and to perceive "air" and space between them. This perception is IMHO a major part of the pleasure of listening to high-fidelity sound reproduction.

But for most music - including, I would argue, any style including classical where the original soundstage is wider than one's listening room - imaging based on a live-performance standard is a fool's errand. In this regard, I'd make a final note that there are landmark rock albums where various pressings over the years and decades have been released with reversed channels. And for some of them, there are so many pressings/versions with each of the two L-R orientations that it's actually not known which is the "correct" channel orientation. So if having the entire soundstage totally backwards is acceptable, then it makes it a bit pointless to go for a "true live" soundstage model.

 

[Cosmik]

The way I see it, imaging is the essence of stereo, and certain types of living room-friendly speakers can be demonstrated that produce compellingly perfect imaging: a spread of musical sources from left to right with the two speakers working together to render the apparent positions seamlessly. It isn't a fragile effect dependent on a super-critical listening position or perfectly symmetrical room.

(Imaging stability *is* recording dependent, and Blumlein stereo or panpotted stereo is the 'correct' way to record stereo for speakers).

So that is the benchmark, and it isn't hard to achieve. For any other result, a lack of imaging could be deliberate: choosing an omnidirectional speaker and listening from a far distance..? Or choosing 'blurred' or mono recordings. Conceivably you could process your stereo deliberately to avoid the imaging.

But if you are hoping for imaging, and using an imaging-capable 'system topology' but not achieving imaging, then maybe the speakers could be improved upon.

 

[RayDunzl]

Having both narrow and wide dispersion speakers in this relatively untreated room... The wide-dispersion speakers are immediately outboard of the narrow dispersion. The positioning difference is about 15 inches of width.

I prefer the narrow dispersion - the image presented is nicely, uh, focused, when listening critically.

The wide dispersion speakers excite many additional higher SPL reflections - the image presented is less, uh, focused.

I have no hard preference when listening casually, and off axis, at moderate levels.

I suspect if the walls and ceiling were deadened, reflections would be muted/eliminated, and the two would present an equivalent image.

Impulse response - showing reflected energy over time.
Red - Wide, Blue - Narrow dispersion, 6 measurements of each. Left, right both, with and without "correction".

 

[Stonetown]

For best audio, I believe loudspeakers with higher directivity is better. We get less reflections from the room walls, and hence we will not hear where the loudspeakers are standing. Using more omni directional speakers, the reflections makes our brain calculating where in the room the sound is originating from so we hear where the speakers are actually standing. With directive speakers it sounds more like listening in headphones, and the stereo image gets better. Drawback is that the audio is good only in a single sweet spot for one person. For a room with many listeners the more omni directional speakers are better of course but with the drawback mentioned before. Very different goals for the optimizations and we have to decide what we want.

 

[JanRSmit]

Where?

See, that's the problem- the signal fed to them is one-dimensional. The soundfield created (and I choose that word deliberately) is 3 dimensional. So by definition, the speaker cannot and will not "reproduce faithfully the electrical signal fed to them."

The point is that we listen with two wars, horizontale placed with soms head in between. Whatsapp we head, ie our ears capture is om essence 2 dimensional. We have no sensors to capture verticale info. Our brains Will however associate noise like airplane noise with height. Imagine an underground plane, not possible.
So we hear 2 independent ears, and our brains pit that together to some soundfield, using alle kinds of experiences, including visualisatie.

 

[SIY]

Yes, we do, but the point remains- the speaker puts out a 3D soundfield, i.e., it is not invariant with any translation in 3-space. This is a response to a 1D electrical signal, thus it is in essence creating new information that does not exist on the recording.

Different speakers in different rooms will create different illusions for different people.

 

[RayDunzl]

New Experiment:

 

[Ron Texas]

@RayDunzl forgive me for asking, but which speaker is wide and which is narrow in this test?

 

[RayDunzl]

Wide - JBL LSR 308

Narrow - Martin Logan reQuest (1998 vintage) - dipole electrostatic panel crossed to sealed woofer (and additional subs) at 180 Hz

The bigger one.

You can buy a pair for $1~$2k from whereever. There's a lot of them around.

 

[watchnerd]

A recent editorial:
https://www.stereophile.com/content/clowns-left-me-jokers-right

A case is made that in except for a few seats image locations are not precise in a symphony. Rather, it is a shapeless monophonic mass. I have noticed this myself although it is more true of typical amplified music played in small venues. What is it we are in search of, and is it the right quest? Does this shed any light on why the overwhelming majority is happy with low data rate lossy recordings and earbuds? Is it the lurid colored neon lights, cheap beer, barmaids in shorts, and hot breeze on Austin's 6th street which makes the music incredibly special? Is what we are craving mostly the magic of some really talented recording engineers?

The editorial does provide various counter arguments.

The logical underpinning of this argument is the fallacy that that a live acoustic performance is the best reference for a recording of the same.

It isn't.

Do the authors think recording engineers use a single set of stereo binaural mics, in a dummy head, placed in seats in a concert hall?

They don't, and stereo recordings (usually made using multi mics) themselves are an artifice, an artistic work unto themselves.

 

[stunta]

New Experiment:

View attachment 23531

Assuming this was measured on-axis, can you repeat the measurements off-axis?

 

[RayDunzl]

Do the authors think recording engineers use a single set of stereo binaural mics, in a dummy head, placed in seats in a concert hall?

They don't, and stereo recordings (usually made using multi mics) themselves are an artifice, an artistic work unto themselves.

--- from The Real Frank Zappa Book - by Frank Zappa, Peter Occhiogrosso

Recording the LSO

Then it was time to record. I tried to forget what had happened the night before, put on my record producer helmet and set out to confute the laws of physics.

The French horns (eight of them) were one of the early problems. We put several Telefunken U-47s on them, but they were useless.

The problem was the percussion (six individual stations) all around in back of them -- we were getting tons of leakage. Even on a tight cardioid pattern, we picked up blobs of low-end mung that was wandering around the floor -- so I said, "Although I've never used a PZM [pressure zone microphone] on a French horn before, what have we got to lose?" We took the large 2½ and put it on the floor behind them and that did it.

We were using a KM-84, flown high over the timpani. We gave up on that and used a 4' X 4' Plexiglas sheet to increase the boundary on one of the PZM plates, put it on the wall behind the timpani and got a great sound.

We were using all the Prescribed Mikes for various instruments but, given our acoustic situation in this nasty room, they just didn't work. For example, we tried two AKG 414s, flown above the conductor's podium, for ambience, but all they gave us was an ugly room sound that was like, "Hey, you want to hear the air conditioner? Let's go. Here it is."

Recording musicians want their instruments to sound (at very least) GLORIOUS -- they've been recording for years and they're used to seeing those gray, heavy, serious-looking microphones. ["Oooh fuck! Look at this! I'm going to sound BITCHEN in this!"] THEN somebody comes in and puts a plastic dome with a PZM in it over their heads and they go: "MY TONE! MY PRECIOUS TONE! I'm going to sound PLASTIC!"

The orchestra already had a strange attitude because everything was different from the way they were used to working. They weren't in a concert hall -- they were on a dead and dusty soundstage. They didn't have the BIG GRAY MIKE -- they had the little plastic bubble over their heads (or they had part of a plastic box on the floor). They had LITTLE TINY WIRES, not BIG SERIOUS WIRES. They could trip over these things (especially since they were so fond of alcohol).

 

[RayDunzl]

Assuming this was measured on-axis, can you repeat the measurements off-axis?

Can, later... gotta do some outdoor housework now...

 

[sergeauckland]

The logical underpinning of this argument is the fallacy that that a live acoustic performance is the best reference for a recording of the same.

It isn't.

Do the authors think recording engineers use a single set of stereo binaural mics, in a dummy head, placed in seats in a concert hall?

They don't, and stereo recordings (usually made using multi mics) themselves are an artifice, an artistic work unto themselves.

Whilst it's true that many if not most even classical recordings are multi-miked, there are still some labels, like Nimbus, that record in a purist fashion. They record in long takes, generally each movement in one take, using the edit, as they put it, "to save a performance, not to create one".

Decca also used to record classical music using the Decca Tree microphone arrangement, which I've used successfully myself for larger ensembles.

Binaural dummy-head recordings are something of a special case, and personally I don;t rate them particularly highly. I like Blumlein crossed figure-of-eight or coincident cardioids for stereo recordings, or indeed Decca Tree. Having said that, for live broadcast of small bands, it tends to be close-miked and pan-potted as drums obliterate everything.

S.

 

[Juhazi]

https://www.stereophile.com/content/clowns-left-me-jokers-right

Actually I agree, I have found that listening to mono pink noise is a very good stereo test!

Ray, I don't think that ML panels are really good example of narrow disp speakers. The backside radiation adds to ambient sound a helluva lot! 15"bass+big horn is what I consider narrow. Best imaging that I have heard came from diy synergy horns that are roughly 1m wide and highpassed around 150hz. It was like having ears of graz lag!