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
#4,486 · (Edited)
Some perspective from my work as a recording engineer.

Pop music recorded in recording studios uses different mic techniques, sometimes using one mic (mono), sometimes two mics in a stereo configuration (like on a drum set or piano). Sometimes multiple mics are used and blended together. A common technique to mic guitar amps is to use a few mics close to the speaker. Additional "room" mic(s) may be used. The final mix of mics is by taste, up to the engineer/artist/producer. You can get a very "in your face" sound or more "amp in a room" sound. Recording studios are chosen for their acoustics, engineers/producers swear by certain studios for how good they sound for drum sets etc. In addition to using the natural sound of the room, artificial reverb and delay is used to create a "sound" that is not so direct and in your face.

Recording schools have students analyze recordings and have them attempt to match the sound of the original recording. Great ear training and research. Knowing what equipment was used, mic techniques, effects processing etc is a great way to learn the skills. Imagine trying to duplicate the sound of a Hendrix song mixed by Eddie Kramer.

Classical music recordings use "natural stereo recording techniques": A/B, X/Y, ORTF, NOS, M-S, Blumlein, Decca Tree etc. The DPA website has an excellent description of these techniques.

https://www.dpamicrophones.com/mic-university?filteroption=408&searchtext=&lowerfilters=

In addition to the main mic arrays, spot mics are frequently employed to help balance issues. The goal of the engineer/producer is to get good instrumental balance and capture the sound of the concert hall and instruments with a good direct to reverberant sound balance. When recording an orchestra, engineers use height and distance of mics (hung from above or on tall mic stands) from the front row of instruments in an attempt to get a good sense of front to back "depth of field". When mixed to stereo, natural imaging is a goal. Instruments should appear in the image like they do on stage, yielding a good phantom image perspective of hard left, left center, center, right center and hard right locations.

The ability of a pair of loudspeakers to reproduce the sound of the recording with full frequency range, dynamic range, low distortion and the ability to create a good stereo image is what I look for. Stereo image includes left to right and front to back perspective. The use of subs can help achieve smooth, extended low frequency reproduction in the room. Varying degrees of toe-in will affect the stereo image and the tonal balance.
 
#4,491 ·
I really wouldn't suggest trying that with speakers that don't have very controlled directivity. By design, you'll be sitting far off axis of the speaker to which you are closest. You want that to cause a lower volume level from that speaker, but the shape of the frequency response needs to be close to that of the on-axis response for decent sound.
 
#4,492 ·
From the article:


For now, assume that the whole spectrum keeps the same general shape ("constant directivity") at different angles off-axis, but gets lower in intensity as you go further off axis (“controlled directivity”). This is usually accomplished with large-ish waveguide-basedspeakers.
;)
Craig
 
#4,493 ·
Exactly. But for some reason many try it with non-suitable speakers and decide it doesn't work. It should also be pointed out the whole idea is to get a better experience for multiple listeners--the main application being home theaters in this context. For a "listening room" with a single chair at the sweetspot, I wouldn't bother.
 
#4,494 ·
Does it do a better job than a hard CC speaker?
 
#4,495 ·
I don't think anybody would claim that. But I think there can still be benefits even when you have a center channel locking that down.



If you're sitting far from center, say, right in front of the left speaker, music from that speaker can be overwhelming and make it more difficult to hear dialog from the center channel and music from the right. With this arrangement, that left speaker is knocked down several db for you, improving how well you can hear the center and right channel which may improve the experience from a movie soundtrack or multi-channel music.
 
#4,527 ·
Often when I read someone's interpretation of the intent of this thread I go back and carefully read the first post, trying to better understand the intent of the OP and not put my own spin on it. After you read the first post a few times little things start to stand out. While some seem to think the OP promised to post all the data, in fact he consistently says "we" (rather than "I") will post data "when available." The fact that "we" was used frequently in the first post suggests this was intended to be a collegial discussion with many participants sharing and learning from each other.

In fact, that's exactly what has happened. Many have contributed to sharing available data demonstrating "What the Science Shows" about speakers. That's what has made the thread so successful and keeps it going strong. The fact that a few are disappointed that the thread hasn't lived up to their expectations should come as no surprise. No forum thread in the history of the internet has ever completely satisfied everyone who read it and this thread is no different. Forum threads die when they outlive their usefulness and the activity in this thread over the past 24 hours alone suggests that there's no need to start thinking about funeral services. :)
 
#4,529 ·
Often when I read someone's interpretation of the intent of this thread I go back and carefully read the first post, trying to better understand the intent of the OP and not put my own spin on it. After you read the first post a few times little things start to stand out. While some seem to think the OP promised to post all the data, in fact he consistently says "we" (rather than "I") will post data "when available." The fact that "we" was used frequently in the first post suggests this was intended to be a collegial discussion with many participants sharing and learning from each other. ...
Mr. Voecks was saying "we," not "let's."

Considering his signature, I assume that when he wrote "we," he was referring to either Harman or Revel or a particular group within Revel.

Either way, it doesn't matter. He's sitting on a mountain of data. It's already "available." If he wanted to post a useful amount of it, he could. The only possible roadblock would be getting corporate permission. The charitable interpretation of the situation is that he couldn't get permission from corporate. The uncharitable interpretation is that he never intended to post a useful amount of data and only really intended to post three spinorama plots that cast his speaker brand in a positive light.

People have apparently been in touch with him to try to get this sorted out. It will be interesting to hear what happened from him directly when he circles back to this. I'm not sure why he hasn't yet. Is he still on vacation?
 
#4,530 ·
#4,531 ·
Not a lot. Interestingly, I had never known of the publication "Technical Acoustics" until now. It is, to say the least, obscure. For a topic so central to sound quality and perception, why not publish in the mainstream journals: Journal of the Audio Engineering Society or Journal of the Acoustical Society of America where it can be peer reviewed by people with relevant research experience?

If one is looking at the audibility of vanishingly small time differences, why do the test in a somewhat reflective room at a listening distance of 4.3 m? An anechoic chamber or headphone simulation would have been appropriate.

Whatever the result, there is the overlaying fact that the amplitude response changes with the delay between two loudspeakers. If we truly are sensitive to time differences in the microsecond range there is absolutely no hope for an intact phantom center or other soundstage images in stereo which incorporate acoustical crosstalk at the ears from the two mono channels, superimposed on head asymmetry and HRTF minutiae, superimposed on head location imprecision - just think, he is arguing that a distance change of 2 mm is audible. Small differences are definitely audible in binaural sounds, but he is arguing similar sensitivity in monaural sounds.

I discuss research in Section 4.8.1 in the 3rd edition that indicates that humans are much more tolerant to time domain "errors" than is argued here. Who is right?
 
#4,550 ·
I certainly understand and am sympathetic to how anyone lacking direct experience with the inner workings of the industrial business world could easily misinterpret certain dynamics. From my own corporate experience (not in the AV industry) I've personally witnessed on numerous occasions how marketing and sales departments have final say over what technical information is made public regardless of what their engineering departments may consider appropriate. Based on what I've seen and experienced I would never blame corporate engineering personnel for being unable to speak freely on a public forum. I appreciate whatever I can get from reading between the lines knowing that any participation by technical people on public forums can put them at risk within their corporate structures.
 
#4,555 ·
Yup I've also had some s**tty corporate experiences, totally understand if corporate isn't on the same page as common sense.

Of course any spinorama measurements done by Harman, and any double-blind testing done by Harman, are going to be corporate confidential by default and employees can't just choose to post them on a message forum without approval from the appropriate departments.

Surely Mr. Voecks would know this and not start a thread in order to post a bunch of stuff that he wasn't actually allowed to post though. So it's still a mystery what happened there.

This whole situation with Harman is kinda dumb aside from that since they went to all the time and effort of making the spinorama an ANSI standard and then ended up doing... well, apparently nothing with it, not publicly. So why make it a standard? Then again, corporations vs. common sense...
 
#4,560 ·
AVKV said:
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.
- That might not be as difficult as you may think.
Agreed. Comparing two closely positioned speakers in that same room means that the room will affect them fairly equally. Neither will sound the way it will sound in your room but it is a fair fight.
 
#4,564 ·
In my limited testing of loudspeakers, I find there is quite a bit of variability between manufacturer specifications versus how the speaker really sounds/measures.

Don't know if this has been posted before, but the ANSI/CEA-2034-A Standard Method of Measurement for In-Home Loudspeakers is available as a free download at:
https://www.cta.tech/Research-Standards/Standards.aspx?search=2034

One needs to create an account, but if you are interested, it is worth the time to register and download the document as it is quite an interesting read. It also provides the basis for estimated in-room response calculation. I.e. providing a usefully accurate preview of how a given loudspeaker might perform in a typical domestic listening room. While the predicted curve is quite accurate, it cannot estimate the effects of room modes below transition/Schroeder. But we know how to take care of that ;)

Note that Revision B is in final draft and will also be available for free download towards the end of the year: https://standards.cta.tech/apps/group_public/project/details.php?project_id=540

Enjoy!
 
#4,573 ·
Good question!
 
#4,583 ·
Here is the spin for the Infinity Reference R162 bookshelf, essentially a poorly promoted low-priced version of the Revel Concerta2 M16. This speaker seems to measure relatively well, especially considering its ridiculously low sale price. I have a pair, and they sound so good that I would have a very difficult time justifying the additional expense for a pair of Revels
Agreed. That is a hell of a sale price. You could get 4 pairs of R162s for the price of the M16s. I wish Infinity wasn't the red-headed stepchild of the Harman group now. They didn't used to be. They had many different lines from low to higher end. One of my first real systems was a full set of Interludes including the sub. It's solid, well-built stuff.

I have the RC263 3-way center and it's a beast.
 
#4,584 ·
If I had to guess, probably a result of the Primus line. They apparently had to absolutely slash prices on those to move them. I think I remember Fry's selling pairs of P363s for $99 during one sale. How ridiculous. I should have bought some. I can't imagine Harman was super-pleased with that situation. Might have mostly killed the brand. I'm surprised they came out with the Reference line at all. Maybe they were mostly done with the engineering for the Reference line by the time they were basically giving away the Primus speakers.
 
#4,585 ·
I haven't heard the Primus, and I don't remember if it was in this thread or another, but someone was making a point that they measured very well, but something was off in the tweeter, iirc, and overall they weren't a very good sounding speaker.

That is certainly not the case with the current Reference line.
 
#4,588 ·
Yeah I think that was earlier in this thread. I think the Primus P363 specifically was generally lauded as one of the best speakers ever made at any price (and its spinorama plot is excellent) but the other speakers in the line didn't measure or perform nearly as well. They had their fans but also a fair number of people who didn't like them that much (elevated/harsh treble, etc.). Also I remember seeing some models at Fry's and thought they looked pretty cheap and wasn't impressed with the fit-and-finish of the exterior.

I'm very pleased with how my Reference speakers look. Of course they're not piano-black Revels with curved cabinets but I think they still look good.
 
#4,586 ·
It's nice inexpensive speakers that measure well are available. In my experience over time and recently at Harman where I heard the Concerta2, Performa3, PerformaBe and Ultima2 series speakers all in a short period of time, every step up in price produced a noticeable improvement in refinement and sound quality. This is certainly not always the case within other brands or comparisons between brands and price points.

The Concerta2 line of speakers sounded very good to everyone who heard them (they unanimously won a double blind listening test in the MLL). The PerformaBe line sounded better. The Salon2 and the entire Ultima 2 line is amazing.

At each price point, Revel speakers offer exceptional value in my opinion.

It all comes down to your budget and what level of price/performance ratio you feel is right for you.
 
#4,587 ·
It's nice inexpensive speakers that measure well are available.
Indeed. Not long ago, the minimum price for admission was around $1000/pair. Recently, it seems like it has dropped into the $500 range. Granted, you'll take a hit in the aesthetics department, but there are some exceptional performers in that $500-$1000 range, and if you go just a little higher, it really opens up even more.
 
#4,589 ·
Background
In post #4530 @OldMovieNut asked @Floyd Toole his opinion on Professor Kunchur's paper: http://boson.physics.sc.edu/~kunchu...isalignment-of-acoustic-signals---Kunchur.pdf
Dr. Toole in his reply (post #4531 ) suggested headphone simulation as a more appropriate choice to look for audibility of small time differences. well, I took the bait, hook, line, and sinker, and so I set out to do my investigation.


A Critique of Kunchur's Study
I began by constructing the test signals, and there, I noticed, IMO, a fundamental flaw in Kunchur's test methodology. He set up his test by stacking 2 loudspeakers on top of each other, move the top one back and forth, and asked listeners if they detect any differences in the sound. The flaw was in his choice of test signal. He fed the same 7 kHz square wave to both speakers.
The issue is that when you have 2 adjacent sources producing tones of the same frequencies, they interfere with each other. For square waves, the sum will visually show “smearing” and no longer look like a simple square wave (see below, solid line is the sum, dash lines are sources) but the overall signal power will also change due to interference. The area under the curve (which relates to loudness) for the staggered waves is smaller, thus the listeners will hear a reduced volume. When a listener hears a difference, it is not possible that he or she can attribute the difference to a sound loudness change, or to a time “smearing” effect. That, to me, is a major flaw. (Other flaws I can think of include ignoring diffraction effects due to the changing degrees of mismatch between the front baffles of the loudspeakers.)


The Experiment
At first I thought of using 2 sine tones of different frequencies as the test signal. But then as I tell myself I can't be that original – a test this simple must have been done countless times already (and the results known, hint, hint). Thus, to be a little more dramatic, I decided to “smear” a square wave instead. A square wave is a summation of an infinite series of sine waves (see below).



Since I generate the test signals mathematically (using Python), I have to truncate the series to work with a finite number of terms. This also gives me the opportunity to band-limited the signals to below the Nyquist frequency. The test signal I chose is a 2 kHz square wave sampled at 44.1 kHz. I only used the first 5 terms, with the highest (9th) harmonic at 18 kHz. The 11th harmonic at 22 kHz is too close to Nyquist for comfort and should be inaudible anyway. This truncation of the sine series is equivalent to applying the standard sinc reconstruction filter in most DAC's. The band-limiting also takes the DAC reconstruction filter (or the lack of one) out of the equation. The test signal should pass through the reconstruction filter in the DAC with zero or negligible alteration. For a primer on digital audio, see video: https://xiph.org/video/vid2.shtml
Below are the 3 test signals. “square1” is the baseline. “square2” and “square3” are 2 versions “time smeared” differently by applying different sets of delays to the harmonics. The idea is to have 3 signals that look very different “visually” but have identical spectral magnitude plots (phase plots, however, will be different). The “jaggedness” look in the signal plots is due to the 44.1 kHz sampling, which equates to a sample every 22.7 µs (or 22.05 samples per period for the 2 kHz signal), and is of no consequence (see the video mentioned above).
These 3 waveforms are saved to wav files (44.1 kHz/16 bit) and they are attached. Below are plots of each waveform with each of its harmonics separated out. A screenshot of the wav files in Audacity is also shown to confirm that they are indeed what they are supposed to be. I am not hiding the identities of these wav files because anyone can open them in Audacity to find out. I'll not bias anybody with my own opinion and let him/her find out if there is any audible differences between these very different “looking” waveforms.



The wav files and the Python source code to generate the waveforms are in the attached zip file.
 

Attachments

#4,590 ·
Background
In post #4530 @OldMovieNut asked @Floyd Toole his opinion on Professor Kunchur's paper: http://boson.physics.sc.edu/~kunchu...isalignment-of-acoustic-signals---Kunchur.pdf
Dr. Toole in his reply (post #4531 ) suggested headphone simulation as a more appropriate choice to look for audibility of small time differences. well, I took the bait, hook, line, and sinker, and so I set out to do my investigation. ...
Great test, I listened to the files, they all sound identical to me. I didn't bother to load them into an ABX tester but I assume there's not much point. Thanks for doing this!

Another idea that I had while reading the paper was to do a test with headphones where one stereo channel played a sine wave, and the other played the same wave but offset by several microseconds (possibly up to milliseconds, for curiosity's sake)...

I believe that could also test a person's sensitivity to such small temporal differences?
 
#4,593 ·
^ No way in hell did the P362s that I had measure like that. The treble was not neutral.

This was long before I knew about or had the ability to measure, so nothing but my ears at the time.
 
#4,597 ·
I'm not gonna give away the manufacturer, but on a project I was calibrating this week in NYC, it was discovered that one of the surround speakers was completely missing the woofer. The installers didn't notice because the speaker grill was still attached. Guessing that one was tossed in a box at 4:59:59 on a Friday and missed QC !
 
  • Like
Reactions: Trell
#4,603 ·
I've been listening to stereo with my front 3 channels for years now via Neo 6: Music, I think the new version is DTS neural X music, not sure how different that is. I like how it still sort of sounds like stereo but it takes away some of the stereo affect and sounds more like live music. The center is actually quieter than the fronts so it doesn't draw attention to itself, either.
 
#4,601 ·
If most popular music is mastered to sound well on imperfect low to mid quality speakers (not to mention earbuds), shouldn't we listen to them with a typical frequency profile of a low to mid quality speaker?

Better still, albums can come with the frequency response of the speakers they were mastered FOR and we can compensate our systems to that profile.

Just a thought.
 
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