Selecting Speakers Should Be Easy, Part One
What Speakers Am I Most Likely to Like?
One of the most popular questions around here is "What speakers should I buy?"
There are at least two key aspects to this question, which is really at least two questions.^
What sounds good?
What is "enough" output?
This post is about that first question. The second question, about output, is discussed in another post.
I presume people mean: "What speaker sounds best?" or, rather, more specifically, "What speaker will I like the sound of, best?"^^
Sound Quality: What Speakers Am I Most Likely to Like
Speaker measurements and science cannot tell you what speakers to buy.
However, getting lots of humans listening to speakers, and seeing if they agree regarding which sound better, would be a useful data point to help narrow down the list of candidates for audition and purchase. Science can help with that.
You can have thousands of people listen to the same few dozen speakers in the same room -- all set up correctly -- without the listeners and the staff having any knowledge of which speaker is which, what it costs, what technology is being used, and so on.^^^
Then, if a consensus emerges from all those listening sessions regarding which speakers sound better than the other speakers, that might help you focus auditions and listening sessions on those preferred speakers.
The good news? Experts have been running those kinds of listening tests for decades, and the results have been very consistent. The same speakers tend to be preferred by most people, both expert listeners and novices. So, you can look up the results of those tests and audition the speakers that won those listening tests. Since most people prefer those speakers, odds are good you too will prefer those speakers, so why not start your shopping there?
Of course, not every speaker has been tested that way. So can we make use of those tests to help us get a sense of which untested speaker might also be winners?
Could we measure the performance of the speakers in a large number of ways, and see if the speakers that people like all happen to share some particular, measurable characteristics?
That way, instead of only having a list of some specific speakers we have tested, are we also able to predict whether other speakers, that we haven't had everyone listen to, might also be preferred by most listeners?
It turns out, we can measure the speakers people like, and then predict, based on measurements of other speakers, what most people are likely to prefer even if the new speaker wasn't in one of the listening tests!
Researchers have repeatedly validated those predictions, by conducting actual listening tests and comparisons between speakers, and seeing whether the predicted preference closely matches that actual preference of listeners. Spoiler alert: It tracks very very closely.
What does that mean? That means we can accurately predict which speakers most people will prefer -- without even listening to them, simply based on how those speakers measure.
That does not mean you should buy speakers just based on measurements.
But it does mean you are likely to end up with speakers you are more likely to like if you start your shopping with models that measure well (ie, similarly to the speakers that large groups of people prefer). To oversimplify, those speakers that people prefer tend to have flat on axis frequency response, consistent off axis dispersion, and, all other things being equal, have more bass extension. (The latter aspect becomes much less significant in a system with subwoofers, of course.)
So, where do I find these measurements?
Well, the Consumer Electronics Organization created a specification years ago about how to conduct those tests called CEA 2034 (sometimes called a "spinorama" since measurements are taken all around the speaker).
More and more, serious speaker manufacturers are releasing this data.
And there are several online publications that do these measurements on a wide variety of speakers. Audio Science Review, Erin's Audio Corner, and others publish their data and there is a nice compendium of measurements from many sources, here: Spinorama dot org
Amir over at Audio Science Review has some good videos about how to use these data, as does Erin over at Erin's Audio Corner (and his YouTube channel is the most prolific and detailed on this topic).
The TLDR summary of how to read this data?
Given the same bass extension, the more linear (not flat, but something in a straight line) the SOUND POWER curve is, the more likely the speaker will be preferred by most listeners. (This line is a combination of factors like on and off axis response…..the two things that matter most in speaker preference testing.)
Appendix
There is a very rich and detailed history of how these experiments were conducted, what all the particular measurements can be used for when choosing speakers for theaters and when designing rooms. This post does not presume to be definitive. Rather, this post is a primer to help you make use of this science. Some additional comments to think about if you are so inclined.
1. Early in all this testing, it became apparent that people had an easier time differentiating between speakers if they heard them in mono -- a single speaker. Conversely, the more speakers that were added to a room, the less good humans were at differentiating among similar quality speakers. So sometimes what looks like a significant difference is actually much less significant in real life. That's good. There is less concern with making an error when there is more margin for similar results.
2. All the speakers tested (and predicted) were heard without any EQ being applied. But in most modern home cinemas, we are all using room EQ, bass EQ, and so on. How does that matter? It means that linear response is less important that consistent dispersion....because linearity can be improved through EQ as long as the off axis response (consistent dispersion) matches the on axis response. So if for budget, output, form factor, or another reason you need to cut corners, you can still get great results from speakers with consistent dispersion even if they aren't linear, since EQ can fix the latter as long as the former is already true
3. If you are worried getting speakers for music versus speakers for movies, don't worry too much. If you get speakers with good directivity and are willing to EQ them (which just every home theater system does) and you are going to use bass management (which just about every home theater system with subwoofer does) then good directivity is what you need. But if you are a music purist and won't use EQ and or won't use a subwoofer, then you have fewer options to choose from. More info:
Footnotes
^The two questions:
FIRST QUESTION
So, to the question: Is there a way to figure out what speakers most people like the sound of? Yes, there is a scientific, reliable way to predict which speakers most people will like.
This does not mean that science alone can tell you what speaker to buy.
But science can tell you which speakers most people prefer. That data can help prioritize which speakers to consider and audition first. Why not start with the speakers that 4 out of 5 people prefer in a blind listening test? Odds are (literally) you will prefer them, too -- and there are quite a few to choose from.
SECOND QUESTION
But be sure to ALSO answer the question: What will work in my room? Some speakers play louder than others, so your seating distance can influence which speakers are appropriate for your theater. This is particularly important if you have multiple rows of seating.
^^This seems like a great question to ask in public forum, where you might find some consensus about what other people like to help guide your own purchasing decisions, or a guru who can predict your likes and dislikes without knowing you.
But you will usually get a handful of loud voices with strong opinions that may not be based on fair (or any) comparisons, may not represent what other people are most likely to like, and may not take into account your situation (how far you sit from your speakers, whether you are using bass management, and so on).
Years ago there were studies published by the Audio Engineering Society showing that most people in a double blind test prefer the sound of speakers with flat anechoic on axis response and consistent off axis response.
That was groundbreaking, because they found that different people (age, gender, culture, race, even listening expertise) actually tend to rank order what they hear in blind listening test the same and we can predict those preferences based on how a speaker measures.
So while your tastes may not be like most people, odds are (literally) your preferences are similar to most other people.
It makes sense to prioritize your listening and shopping based on what you are probably going to like more. And you can figure this out by looking at objective data, to narrow down your world of options.
The web site spinorama.org has data about many of these speakers. You can see how accurate (neutral, or flat) each one is, and how it’s dispersion (off axis response) looks. I would try the most neutral consistent off axis speakers first (if you want to do some test listening) and then move on to others from there, if that wasn’t pleasing to you.
If the manufacturer and third party tests don’t tell you the response data of the speaker I would prioritize it lower than the rest to audition.
^^^So called "double blind" listening tests in which even the person running the test doesn't know which speaker is which.