Updated: Oct 30, 2019
Speak with anyone who's had a little experience with recording, or read any audio forum, and there will be much talk about how that microphone is crap and this other is fantastic.
Beginners will carefully and timidly ask for opinions, while old timers will instantaneously recommend this or this other model. Judgements are quick, brash and final.
It must be hard to be a microphone.
As for much else in the world, strongly held opinions not always have equally strong foundations. Let's have a look on how to evaluate a microphone.
A microphone has one job: to capture the sound in the space you place it, and transform the information into an audio signal (that is, "changes in AC voltage over time"). No big news here.
In general, we can say that the more faithful the transformation, the better the microphone.
However, we don't listen to AC voltages.
What we listen to, is the sound emitted from a speaker reproducing the signal originally captured by the microphone - and, usually, after being processed in a number of ways (at least, preamplified).
To evaluate a mic, therefore, we need to make sure we know which sound we're trying to capture. And then use a recording and playback chain which does not affect the signal too much.
This where the first trap lays, because doing that isn't as straightforward as it sounds:
first thing, we need to have a very good sounding room (i.e. one with not too many reflections). If we don't, a great mic will end up capturing very well a very lousy sound - and we may mistakenly conclude that the result is due the mic, while it's the room that is the culprit.
then, we need a relatively clean recording and playback chain, with appropriate gain structure to maintain the signal integrity throughout. Otherwise, along the way, we "ruin" the sound that was captured by the mic, and again we may think that the poor quality of the result is due to the mic, while it's us.
For the same reason, we need flat monitors with good detail, which do not unduly alter the timbre of the sound for the better or worse. Otherwise we will end up "hearing" the monitors, not the mic.
Finally, the playback room must also be good - as described in a good sounding mixing room, we want the listening location to be as dry as possible, so that we don't mistake the sound of the room for the sound of the mic.
Basically, if you take a microphone, you place it in an untreated room and then proceed to record... and/or you listen to the result with bad monitoring or in a bad listening space... you won't be able to say much about its sound.
Which of course doesn't stop people to do just that.
But now you know, so no excuses.
However, let's assume that you do all of the above. How can we say that a mic is good?
Well, of course the basic way is simply listening. If you like the sound of the result, the mic is good for you. Simple as that!
And indeed, no technical spec in the world can replace a simple live test (under the conditions above): in order to be judged, a mic needs to be heard, possibly with more than one source and in good recording and listening conditions.
That given, we still say that certain mics sound better than others. How comes?
The most obvious possible reasons are also the ones less worth talking about, because anyone who has recorded a bit knows about them, and they don't really mean that much.
For example, a perfect microphone would maybe have a perfectly flat frequency response. But often we like microphones who "flatter" the source a little, making it sounds better than it is for real. At least I know I do, when I record my own voice! And it's a common experience that a cheap mic with, say, a hyped 8 K can be perfect for a certain voice and super-harsh for another! So "perfect reproduction" is not really a measure of greatness. Hyped mics will have less general applications and suit certain sources better than others, and that's about it.
About sensitivity and headroom: some mics are sensitive, some aren't - and for most studio applications that's basically all there is to say. A SM7B is fairly weak when it comes to sensitivity, and yet is regarded as a very good mic, and with good reason (I'll come to it in a minute). And yes, mics with more headroom are easier to use... but getting a good sound from them is more about the skill of the user to select and place them, than the mic itself.
Also things like proximity effect and resistance to plosives are more a matter of choice than a measure of quality. Bass level increases due to proximity are often useful and expected by, say, vocalists; and while a mic more prone to plosives can be thought as worse than one that isn't, in reality a bit of proper technique and positioning (not to mention pop filters) allows almost any mic to be used properly.
Handling noise can be an issue, but it really depends on the application, and it's not a big issue for studio microphones, which aren't supposed to be hand-held when recording.
Stuff like construction, manufacturing precision and so forth are obviously important, but you certainly don't need me to tell you that a mic with shoddy soldering and flimsy connector is not as good as one which is better built.
Noise is different. Obviously a mic with lower noise can be considered better than on with higher, so long the underlying technology is similar. Signal/Noise ratio is very important to get a good recording. However, unless it's really bad, a little noise doesn't make a microphone necessarily bad. For example, certain designs can be a bit more noisy than others, but within reason, it's fairly irrelevant. A classic Neumann U47 tube mic has 85 dBA of S/N ratio and thus about 9 dB of self-noise (self noise is usually measured at 1 Pascal SPL, that is 94 dB SPL), and the solid state U47 FET, whose S/N is rated at 76 dBA and therefore has about -18 dB of self noise, but nobody would claim either is a bad mic.
So why splurging a significantly bigger amount of dosh for a Neumann, or Peluso, or Royer, or AEA and so on, instead of that $10 BlingBlong USB mic?
The most significant factor is off-axis response (also called side response), which is a consequence of the mic's polar pattern. This is where a mic truly shines - or not.
That's because polar patterns are created by designing a labyrinth into a capsule, where air (aka sound) is forced into, so that delays and cancellations occur on purpose inside the capsule.
Basically, the capsule designer uses destructive interference between identical sound waves to cancel sound coming from certain directions (this is why it's hard to create polar patterns which are stable with frequency, as lower frequencies tend to be far less directional than mid-highs).
This allows the designer of the mic capsule to build basic polar patterns; if needed, different capsules can be joined to create more complex patterns, or stereo mics and so on.
Now, it's relatively easy to build a capsule that picks up sound well "from the front", but it's much harder to make it so that the off-axis sound (aka coming from the sides) is also picked up (and progressively attenuated) without damaging the timbre.
That's exactly because attenuation is the result of interference created by a physical labyrinth in the capsule, and it's very hard to build a labyrinth that can control the pattern of interference over both a continuous frequency range and a wide range of angles.
As we saw, this has an effect on how wide is the frequency range where the polar pattern shape stays consistent (in turn, that often is key to determine what application a mic is best suited to). But the most critical consequence of the labyrinth is on how it shapes the off-axis response.
Beyond the capsule, the quality of the off-axis response depends on both the capsule labyrinth and the reflective properties of the protective mesh around it, which needs to be designed to limit filtering between the mesh and the capsule diaphragm.
All this means that, especially among non-omni mics, the better mics have invariably better pickup of off-axis sounds - in the sense that the timbre of the sound coming from the sides is not too much altered, and it's in phase with the one coming from the front. They don't color the side sound very much.
"Bad" mics tend to change the timbre much more on sound off-axis sound, may suffer from filtering between sounds comings from different angles, or suffer from internal reflections due to a mesh which is put together without much thought.
So to evaluate a microphone - once you've set it up in a good sounding room and made sure you have good monitoring - you've got to record something (for which the mic is suitable) which has both a front and side component, and listen for the timbre of the the side sound and how it interacts with the front sound.
A great mic will have a great side response.
Before concluding, it's worth to say again that, with a microphone, the technical aspects bring you only so far. It is essential to listen to the result to make a final judgement, and evaluate the mic for the type of usage and source you're dealing with. Therefore you need a good sounding room to record and a good monitoring environment (room + playback system) to evaluate what you hear.
But that given, you'll consistently find that, everything else being equal, what are considered the best microphones tend to have the best (less colored) off-axis response.
A couple final notes.
From a manufacturer's point of view: measuring a mic response is far from trivial and it's fairly expensive. Without going into details (it'd take pages), a couple of important issue is that in order to measure you need constant source sounds (which means, recorded) and they are emitted by a loudspeaker - which in turn must be extremely flat (actually several loudspeakers, since there's no single loudspeaker that can reproduce the entire audio spectrum in a flat manner) and kept far away from the mic - the longer the wavelength to measure, the farther. And since reflections would make any measurement meaningless, the mic manufacturer need to take its measurements in anechoic chambers, which are hard and expensive to build. In particular, due to how sound waves behave, lower frequencies have long wavelengths, and the consequence of that is that, in order to test the bottom response of a mic, you need a seriously big chamber! Not your run of the mill office room for sure.
Also, the manufacturing involved in producing the air labyrinth which creates the polar pattern is often far from trivial - we're talking of very small and precise holes with very precise spatial and distance relationship with each other Production tolerances need to be pretty high, which as usual increases the cost due to a possibly large amount of rejects.
That so many reasonably good microphones can be had nowadays for an affordable amount of money, is one more testament of the increased manufacturing capabilities in the industry brought about by technology.
A well known example of a great capsule that was apparently very hard to fabricate was the AKG "brass"-ringed CK12 capsule, used in the original C12 microphones in the early 1950s. It was so hard to make that AKG soon switched to a different material for the ring - teflon - to be able to continue selling it a reasonable price.