AVS Forum banner

How to Choose a Loudspeaker -- What the Science Shows

550K views 6K replies 267 participants last post by  Rex Anderson 
#1 ·
Choosing a loudspeaker may be the biggest challenge for music and home theater lovers. There are countless brands from which to choose, and even more claims and counter-claims. Since the room has such a profound impact on the sound of a loudspeaker at lower frequencies, and it is impossible to listen in a blind test at an audio store, if they can find one, there is little that an audiophile can do to make a rational decision. Fortunately, science has come to the rescue with a set of measurements that have been proven to demonstrate an extremely close correlation with sound quality, as based on carefully controlled double-blind listening tests. This group of measurements have been adopted as the industry standard for measuring loudspeakers, as ANSI/CEA-2034-A. https://standards.cta.tech/apps/group_public/project/details.php?project_id=165

Contradicting the oft-repeated claim that choosing a loudspeaker is a very personal choice, research has proven that regardless of age, culture, or listening experience, all people with nominally normal hearing generally agree on which speakers sound better than others. Indeed, there is a universal definition of what sounds good. http://www.aes.org/e-lib/browse.cfm?elib=12794 and https://secure.aes.org/forum/pubs/conventions/?elib=12847

In this thread, we will publish the results of these measurements. In addition, we will discuss their correlation to double-blind listening tests, http://seanolive.blogspot.com/2008/12/part-3-relationship-between-loudspeaker.html as well as publishing the results of formal listening tests, when available. We will add measurement results as they become available. The intention of this thread is for it to be reality-based, and to inform and discuss loudspeaker measurements and listening tests. The papers that really started it all are now available for free from the Audio Engineering Society here: http://www.aes.org/e-lib/browse.cfm?elib=5276 and here: http://www.aes.org/e-lib/browse.cfm?elib=5270
 
See less See more
#4,825 ·
Just to be sure I'm following things here. It seems then that although any FR aberration is a resonance, the important thing about the spinorama is it can identify "resonances" with varying degrees of directivity. For example, a cabinet wall "resonance" would radiate in an omnipolar fashion - and visible throughout the curves, whereas a "resonance" caused by a surface defect on the face of the speaker might radiate within a narrower directivity pattern - perhaps only visible in one or two curves. They're both resonances, but not equal for the end user since after the design is complete the end user's only fix is electronic EQ, and the former seems far less amenable to correction than the latter with that particular tool.

I wouldn't have called both of them a resonance prior to this discussion, but ultimately it's just semantics. Hopefully I'm not entirely confused.
 
#4,826 ·
I wouldn't have called both of them a resonance prior to this discussion, but ultimately it's just semantics. Hopefully I'm not entirely confused.
That's exactly how I feel. :eek:
 
#4,836 · (Edited)
The point is that above a certain cycle point, many of the peaks and dips that can be seen (as measured), are not what we perceive. Sure, using time-gating can help - but at the expense of resolution. It's easier, and a lot more reliable, to simply get a decent loudspeaker and leave it as-is. Or as pointed out, buy a $100 microphone, spend some time learning to use free software and how to get reliable measurements, and do it yourself.

I installed a sound system for a gymnasium recently. They did not have a big budget. We bought inexpensive passive loudspeakers with good directivity traits, bought some inexpensive amplifiers with DSP - spent a couple of hours measuring speakers (making quasi-anechoic measurements in a large empty gymnasium) and making a filter to provide a flat (+/- 1.5dB) listening window. A sub in each of the 4 corners provides an even low across the floor. Everyone is happy with the performance, for less than $2k including 8 speakers, 4 subs, 2 amps and a mixer. The science works.
 
#4,838 ·
The point is that above a certain cycle point, many of the peaks and dips that can be seen (as measured), are not what we perceive. Sure, using time-gating can help - but at the expense of resolution. It's easier, and a lot more reliable, to simply get a decent loudspeaker and leave it as-is. Or as pointed out, buy a $100 microphone, spend some time learning to use free software and how to get reliable measurements, and do it yourself.

I installed a sound system for a gymnasium recently. They did not have a big budget. We bought inexpensive passive loudspeakers with good directivity traits, bought some inexpensive amplifiers with DSP - spent a couple of hours measuring speakers (making quasi-anechoic measurements in a large empty gymnasium) and making a filter to provide a flat (+/- 1.5dB) listening window. A sub in each of the 4 corners provides an even low across the floor. Everyone is happy with the performance, for less than $2k including 8 speakers, 4 subs, 2 amps and a mixer. The science works.

Wow, that's a low budget system for that amount of equipment in a gymnasium. With 8 main speakers I assume you have good coverage. I wish more venues would use four subs, one in each corner. Question is, how big is the gym and how loud can the system play? Is the PA for voice only or does it have to play music too?
 
#4,841 ·
The question is: what will degrade the sound more - correcting a possibly (very) local effect for one listener that may or may not be audible, or adjusting the (hopefully correct) direct sound -the first sound to arrive at anyone's ears-.

There is actually a section about this in Sound Reproduction and multiple references throughout the book.
 
#4,845 ·
Indeed. The dip is an issue with most two ways around the crossover, but being 'normal' doesn't make it right.

On the other hand it is probably not an audible problem. Many speakers have such a dip designed into their response because people like it.
 
#4,846 ·
Perhaps I wasn’t clear. The dip is not there in the listening window response, or even the horizontal off-axis response. The measurement you see is steady-state, meaning it also captures all sorts of reflections, including vertical ones (hence the dip being normal, unavoidable even in vertically oriented multi-driver loudspeaker). The drop-off after 10Khz: the tweeter becomes highly directional – since this is avery large space, the curve you see should quite closely follow the sound power response. Again normal behavior under these conditions.
 
#4,847 ·
I've never set up a system in a large space like a gymnasium so I don't know much about it. Your FR graph looks quite good. Nonetheless, I have some questions:

You have 8 speakers and your measurements are of 2 speakers, correct? What signal is sent to the other 6 speakers? Are you using "stereo" with 4 "Lefts" and 4 "Rights", or some sort of 8-channel, "surround sound" system, or are the same, mono, signals sent to all speakers? If all the speakers get the same signal, why not measure all the speakers at once?

In a large, "echo-y" space like a gym, with all hard reflective surfaces, I would think the time-gated response would be quite different than the un-gated response, no? Was this a gated or un-gated measurement?

What smoothing did you use in the measurement?

How do you "time" the subwoofers relative to the speakers so as to avoid phase errors around the crossovers at different listening positions? Does the response around the speaker/subwoofer crossover look the same at multiple measurement points, or does using 4 displaced subwoofers eliminate that from consideration? Was the same, mono, signal sent to all the subs, or was each sub "timed" independently? I presume you EQ'd the subs as one combined sub, correct?

Sorry for all the questions, but I'm just trying to understand how to set up and optimize a system like this. Clearly, it's different than a small room HT. Thanks.

Craig
 
#4,857 ·
2 on each wall / stereo. If you were to face one wall, the four speakers to your left would be L and the four to the right R. As said this will be mainly used for modern dance practice.

edit: I tried to make a little layout in text, but it doesn't seem to show as it should here

I could have gone mono, but opted for stereo for the satelites. To be completely honest. I have no background in audio apart from a healthy interest.
This project came into my hands because there was little money left for sound, and I said I would be able to get good sound in that room for the available budget if they let me handle it. Honestly, it came out as good as I could have hoped for. I just followed the guidelines as portrayed in Sound Reproduction.

Most of the discussions about the subject here are what I consider nitpicking. But the fact is that simply 'following the book' plus learning how to get decent measurements out of a loudspeaker, and understanding what can be corrected, provided a very nice result.
 
#4,864 ·
That dip is a classic example of something that shouldn't be corrected. Even with spinorama measurements, it seems it can be sometimes difficult for people to grasp 'cause and effect'.

With any vertically oriented multi-driver system, at some point (especially as the frequency goes up - and the wavelengths get shorter) under specific vertical angles, drivers in a system, optimised for listening 'on-axis' will simply no longer be in phase, causing a suckout where they overlap. The good news as explained by F. Toole, is that we are simply less sensitive to these vertical reflections, as our ears are in the wrong plane to perceive them very well. So as long as the direct sound field is flat, and the early horizontal reflections are of similarly good quality, it would quite literally degrade a loudspeaker by correcting such 'anomalies'.
 
#4,867 ·
It really depends. These systems are designed for the masses - who don't have speakers that measure that great, and non-optimal rooms - so they perceive very audible benefits. Black and white rules need not apply.

The game is different if you have really good speakers in a well-thought out room.
 
#4,879 ·
I'm not an expert, maybe this answer is too simple for what you're looking for, but all other things being equal, there's a preference for louder sound/speakers. That's why you have to take care to even out the volume when evaluating speakers, which may mean turning your amplifier up or down depending on the sensitivity of the particular speakers you're listening to. This is called level matching.
 
#4,898 ·
Every now and then we got money for new monitors and were able to get some in for listening tests. When I was doing the speaker comparisons, I tried to get them pretty well level matched with pink noise and program material using an SPL meter. It just puts them on a more level playing field. Back in those days, I was not attempting to do blind listening tests, but I did figure out it was OK to use one speaker in mono instead of two in stereo.
 
#4,903 ·
I have a question that @Floyd Toole and/or @SoundnWine could offer some insight to. I've been digging into some of the earlier listening studies regarding the mono/stereo tests and the Regression model for predicting loudspeaker preference and noticed a possible correlation between the high weightings of the On-axis sound and the way the tests are performed. I noticed in all of the setups, the speakers are pointed directly at the listener. Here is a snip of the early setups, I believe now the shuffler puts the speaker directly up front and center aimed straight at the listener.



Looking at the preference models, I saw for the Test 1 model the On-axis parameters were responsible for 45% of the weightings of the overall model while it's still 31.5% of the weighting in the generalized model. The Mono/Stereo listening study showed me that not only will the same speaker be preferred in Mono and stereo but also that placing a speaker in the front in Mono isn't going to change the preference compared to a more typical placement in the corner.

My question is do you believe the way the speakers are placed directly on-axis to the listener is going to bias the test and/or the preference model towards the On-axis sound relative to a more typical setup? By typical setup I mean the speaker being about 15 deg off-axis to the listener and closer to the side walls so that now the direct sound heard is actually closer to the listening window response and presumably more reflected sounds are now heard by the listener. I tried to do my homework as best I could but I am curious about your thoughts on the subject, thank you.
 

Attachments

#4,905 ·
The key factor is that the first-arrival direct sound has a special place in the heirarchy of sounds arriving at a listener. You are asking, I think, if it matters whether that direct sound is the on-axis or 15 deg off axis sound delivered by the loudspeaker. The answer has to be dependent on how well the speaker was designed. In today's best products there is very little difference between the on-axis frequency response and that of the listening window that includes components up to 30 deg off axis. In such cases it very likely does not matter. In other designs there can be large differences, and these are problems - products to be avoided.

It is generally good practice to aim the speakers at the listeners because in most cases the on-axis sound has received special attention in the design process. Of course, the amount of sound reflected from the adjacent side wall is a factor, but this is a far off-axis, significantly attenuated component, especially if the speakers are aimed towards the prime listener. Again, what is heard will depend on the specifics of the speaker design.
 
#4,904 ·
Advice - deployment of side speakers

If a momentary change in topic is ok??

I have a pair of SVS Ultra towers (can't find any spin data), two subs, two Ultra Surrounds, and four SVS Prime elevation speakers. My room is abt 16' long, 8' hi & 11' wide. Given the narrow room, would it be better to employ the Surrounds as dipole side speakers, to minimize direct blasting into one ear of the listeners, and to use two of the Elevation speakers as left and right back surrounds, and the other two Elevation speakers as side ceiling speakers? Or put the Surrounds in back, or replace them with more direct-facing back surround speakers? I'm on my third reading of Dr. Toole's book, there's a lot to learn! I greatly appreciate the assembled expertise in this forum. Thanks again.
Jack
 
#4,906 · (Edited)
If a momentary change in topic is ok??

I have a pair of SVS Ultra towers (can't find any spin data), two subs, two Ultra Surrounds, and four SVS Prime elevation speakers. My room is abt 16' long, 8' hi & 11' wide. Given the narrow room, would it be better to employ the Surrounds as dipole side speakers, to minimize direct blasting into one ear of the listeners, and to use two of the Elevation speakers as left and right back surrounds, and the other two Elevation speakers as side ceiling speakers? Or put the Surrounds in back, or replace them with more direct-facing back surround speakers? I'm on my third reading of Dr. Toole's book, there's a lot to learn! I greatly appreciate the assembled expertise in this forum. Thanks again.
Jack
I'm hoping that you made a typo when you wrote "dipole". If you indeed have read my book, Section15.8.3 shows the gross flaws in dipole (bidirectional-out-of-phase) surround loudspeakers. These are not to be used in any modern multichannel system, especially immersive systems, where all loudspeakers are required to deliver high quality, neutral, direct sound to all listeners.

As illustrated in that section Bipole (bidirectional-in-phase) loudspeakers are very different devices, capable of delivering excellent sound quality over the very wide dispersion angles required in small/narrow rooms with multiple rows (Figure 15.9).

Figure 15.8 shows suggested layouts in which the "side" surround speakers are placed forward of the prime listener. This alleviates some of the attention-getting problem of those speakers in narrow rooms, and at the same time improves the perception of envelopment which these speakers are primarily intended to deliver. The rear speakers deliver the fly-over illusions. The notion that side speakers need to be at 90 deg is based on a tradition beginning with the original 4 channel Dolby surround system - they simply haven't changed, even though there is good reason to do so. Don't confuse the customer.

Think of what is done in "reference" cinemas: there are "side surround" loudspeakers distributed along the entire side walls.
 
#4,910 ·
Thank you Dr. Toole for your reply. Yes, it was a typo and I meant to say bipole (a typo apparently duplicated in the second paragraph of your response:). It was Section 15.8 that made me think of using the "Ultra Surrounds" as side speakers. Given that they are bipoles, should I still locate them forward from the main listening position, or would 90 degrees be better here?

Unrelated (kind of): High-frequency hearing loss, tinnitus (for me a high-pitched ringing sound), and inconsistent judgement of sound quality, all affect my enjoyment of music...I even sometimes sense distortion (IM, perhaps) that isn't really there, in live symphony performances! Growing old is truly not for sissies. Thank you again for your invaluable contributions to the body of knowledge.
 
#4,912 ·
Ha Ha - those damn spell "correctors" get all of us, don't they. :(

I would still put the B "Bee" ipole (I just had an argument with my spell checker) speakers forward of the listening position. Studies have shown that the optimum angle of incidence for maximum envelopment is +/- 60 deg - coincidentally the angle of arrival of side wall reflections in concert halls. Envelopment turns out to be the dominant factor in assessing the quality of concert halls. I experimented with the angle for my first home theater back in 1988 and ended up with the speakers around that angle - it was a music priority system.

However, it was also a 7-channel system - before its time - so I had rears as well. A Lexicon CP-1 (Consumer Product number One) offered a multitude of opportunities to experiment and I did. If at all possible it is advisable to have 7 channels. As shown in my book I had a large "classical" music listening room as well and it was interesting to compare the auditory impressions of the two. The latter room ended up with bipole Mirage M1s generating a very spacious soundstage in the large room. Figures 7.18 and 7.19 in the 3rd edition show the two very different rooms.

Now I have a 9.4.6 channel system in a very different room in California.
 
#4,913 ·
Dolby spec is 90-110 degrees:

View attachment 2616052

I find I get better envelopment when they are more than 90 degrees (slightly behind) from me. YMMV.
Interesting, and not a problem if you like it. I really think that a 10 deg difference at the side is mighty small - well within localization blur. A 30 deg difference would be more audible. It is interesting that they allow an 80 deg (30 to 110) empty angular window from the fronts to the sides when in real life listening the most important sounds arrive from the frontal hemisphere, including those highly important lateral reflections in concert halls. I have long felt that Dolby had some 'splainin' to do, but they rule the world, so who am I to differ? They are in business, I'm not.

Related to this is the interesting findings shown in section 15.7.1 in the 3rd edition where simplified speaker arrangements are compared to much more elaborate ones (12 and 24 channels) in delivering envelopment. It turns out that the conventional 5.1 arrangement works well, as do several others. Just avoid symmetrical front/back angles. It is important to note that adding a center channel improves envelopment (Figure 15.7). Why? Because in real life there is front arriving direct sound against which to compare later arriving reflections. Stereo has no direct frontal direct sound - it is a phantom image. Yet another compromise presented by stereo.
 
#4,919 ·
I have been running a phantom center since March...I reasoned my speakers image so well in stereo (solid center image giving the impression a center speaker was used) that I decided to forgo a center speaker altogether. I've recently felt something was missing tho. I know why now. Fascinating reading...thanks to all contributing.
 
#4,920 ·
how does "How does electromagnetic induction work in a large audio speaker?"

no idea how this works but have heard this induction can play havoc???
 
#4,926 ·
if my system plays a speaker setup say with klipsch towers and gives x results...then same system plays focal towers set way, then plays magnepans set way...I think all would agree they all sound different. no book will tell you which sounds best to you. but everyone will have opinion on best in my room and with what music. If I cared to share simplistic measurements wouldnt change anyones mind. the point of sharing 3 different types of speakers in the same room just shows people like different stuff.
 
#4,931 · (Edited)
Thanks, but what I meant, as before is a "raw" matrix.
So basically "inside" of a mixer, so I can implement it myself (maybe if I can see a benefit).
Some kind of a table that contains gains and processing derived from 2 channels to 3 channels.

I can, of course, play with m/s processing myself to try achieve something, but a "starting point" would be nice.


I've found a plug-in that can do that (auro), but I am not sure if it is possible to use in Linux
 
#4,947 ·
Thanks, but what I meant, as before is a "raw" matrix.
So basically "inside" of a mixer, so I can implement it myself (maybe if I can see a benefit).
Some kind of a table that contains gains and processing derived from 2 channels to 3 channels.

I can, of course, play with m/s processing myself to try achieve something, but a "starting point" would be nice.
I see. It sounds like you're trying to develop your own upmixer. I can't help you with that. :eek:
 
#4,942 ·
Compromised music videos

I happen to enjoy a good video concert, as do most of my friends. Seeing the performers and hearing decent sound is high entertainment. It is also one of the few opportunities to hear multichannel audio for popular music, and artists we appreciate.

Sadly, what we get is often either no center channel or what should be in the center channel distributed among all three fronts. Mostly we get a conventional stereo soundstage, with - maybe - some acoustical ambient sound, and - probably - some surrounding applause at times.

Several years ago, at an AES convention, there was a workshop demonstrating what surround sound could do. On stage were several well-known pop and country recording engineers, and one from NHK the Japanese national broadcast network. I attended a bit of the setup and heard samples of what was to be demonstrated, and during the event I sat adjacent to the center aisle - interesting that at a "spatial sound" event, there were no symmetrical stereo seats - one could stand in the aisle, as I did from time to time. It was interesting that few people did.

What did I hear? From the pop and country guys it was old fashioned stereo with some ambiance mixed in - no noticeable center channel. For me, about 3 feet away from the symmetrical axis the center phantom images flopped almost completely to one side - surprise, surprise. Much discussion centered around the "difficulty" of mixing a center channel and how one could not trust customers to have one, much less a good sounding one. At the time, even Bose was delivering identical LCRs :). All the demos were pan-potted pop.

Then it was NHKs turn. It was explained that they faced the challenge of using a center channel in a way that did not disrupt the complex, natural, soundstage of classical music but that delivered high quality sound and anchored the central images for off axis listeners. They succeeded remarkably well. One could get up and walk around in what amounted to a generous "sweet spot". It was impressive. This was, and I presume still is, what Japanese listeners get off air. Lucky them. It is also what is in numerous well-recorded multichannel programs.

So, if the engineers are not educated in the use of a center channel and choose to blow off anyone who has a decent - not even superb - sound system, this is what we get. I always get up and listen to what comes out of the center channel in concert videos. I am routinely disappointed. Fortunately the music survives - "translates" as it is called - but it is not a "literal" translation. Having a good, well edited, accompanying video is significant compensation.
 
#4,946 ·
...

Then it was NHKs turn. It was explained that they faced the challenge of using a center channel in a way that did not disrupt the complex, natural, soundstage of classical music but that delivered high quality sound and anchored the central images for off axis listeners. They succeeded remarkably well. One could get up and walk around in what amounted to a generous "sweet spot". It was impressive. This was, and I presume still is, what Japanese listeners get off air. Lucky them. It is also what is in numerous well-recorded multichannel programs.

So, if the engineers are not educated in the use of a center channel and choose to blow off anyone who has a decent - not even superb - sound system, this is what we get. I always get up and listen to what comes out of the center channel in concert videos. I am routinely disappointed. Fortunately the music survives - "translates" as it is called - but it is not a "literal" translation. Having a good, well edited, accompanying video is significant compensation.
So we're talking discrete three channel. Any thoughts on using Dolby Pro Logic II Music, Dolby surround, or NAD EARS to extract the center channel? Particularly for classical music?
 
#4,944 ·
It's taking a long time for the music recording world to catch up to the surround audio for video world. Why? Cost! Recording studios have been working in stereo for decades. The cost to set up control rooms to mix in even 5.1 is an enormous undertaking. Most mixing consoles are still designed to mix and monitor in stereo. Major retool to design them for multichannel. Big name manufacturers (Neve, SSL, API) that make the high channel count consoles in many of the remaining big studios are vintage pieces of equipment and are worth many hundreds of thousands of dollars. Control rooms have been designed for stereo. Major issue to change to surround, it's not just add some more speakers. Many control rooms I worked in were barely big enough for two speakers, the console and the racks of outboard equipment necessary to make records. It's not just recording studios, you also have mastering studios that have been designed and built for stereo. The demand for multichannel music mixes is growing and I'm sure there is some progress on that front, but it's not going to happen for a lot of music only releases any time soon.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top