For ages, I’ve been frustrated by the lack of real-world information about the very expensive tools we use. Before I got into sound, I used to do reviews of computer hardware, and it was unthinkable to me that I would make a purchase without knowing what I was buying. In film sound … there just aren’t very many people doing what we do, so we tend to make our decisions based on what everyone else is using. Not many of us get to do comparison tests before we lay down a large amount of money, especially outside the major film centres.

I’ve been on a mic acquisition kick lately, and I’ve ended up with too many shotgun mics, so I need to figure out which ones I want to keep as my workhorses. To help me choose, I decided to test to them all under controlled circumstances and see what I could learn. These are the mics I tested:

Sennheiser MKH-60

My workhorse for years. I originally picked it up because it was cheap (used) and available, and because I had enough other Sennheiser gear that I thought I could trust it. I’ve been happy with it, but I’ve always had an inferiority complex about it because it’s not what most other mixers in my area use (see the next two choices). Not a very common choice these days, especially since the MKH-8060 was released. The MKH series has perhaps the lowest self-noise available due to their RF condenser design, and is well known to be durable in all sorts of weather and a wide range of temperatures.

Schoeps CMIT-5U

If there’s one mic that screams professional sound mixer, it’s the CMIT. The Schoeps name has clout, and the CMIT is standard. I know many mixers who use them indoors as well as out, although most seem to have switched to the MiniCMIT due to its smaller size. It has a reputation for not doing well in humidity, though their latest “Generation D” capsule has supposedly fixed this. I don’t have that version.

Sanken CS-3e

I had to borrow this one. In my area, this is even more common than the CMIT for outdoor use (I think most mixers I know have both), due to its multi-capsule, noise-cancelling design. It has a reputation for being a “laser”, and being capable of very high noise-rejection possible without sacrificing on-axis fidelity.

Schoeps SuperCMIT-2U

The CMIT … but with noise cancellation. A different design from the CS-3e, but the same basic idea: Use a second capsule to reject off-axis sound through noise cancellation. It is even more eye-wateringly expensive than the regular CMIT, and is awkward to use because it outputs AES42 digital audio rather than analogue, and it requires 10V digital phantom power, which is not interchangeable with regular 48V phantom power. It uses the same capsule and interference tube as the regular CMIT, and it outputs an unprocessed signal in addition to the noise-reduced version, so in theory you get a CMIT part and parcel with the noise cancellation. I really want to see if that is true.

One warning: My SuperCMIT is not in pristine condition, and it has a 1.5kHz pure tone in the noise floor of both the processed and unprocessed channels that is audible under very quiet conditions. I didn’t hear it in any of the test recordings for this shootout, but Schoeps has confirmed that this isn’t normal and I will be sending it in for service.

There are several other comparable mics I would have a liked to have tested, but don’t have ready access to. Maybe in the future: DPA 4017, Neumann KMR81i, Schoeps MiniCMIT, and Sennheiser MKH-8060. I would also have liked to have tested a Røde NTG-something (don’t know their lineup that well), Rycote’s HC-22, and perhaps the DPA 2017. These are cheaper, “lesser” microphones, and I would like to find out if the more expensive crop is really worth the money. Next time… Last, I wish I still had a working Sennheiser MKH-416, purely because it’s such a common, iconic point of reference.

Methodology

I tested all the mics together, bringing them as close as I could to a single point without creating too much of an acoustic shadow. I ended up mounting them all on a single C-Stand, angled just slightly down at eye level, roughly 10 centimetres apart, and offset diagonally so that the horizontal and vertical planes were clear for all mics. This isn’t especially realistic, since standard boom position is a much steeper angle towards the ground, but it made it easier to test off-axis noise, as I could walk around behind them and be reasonably close to 180° off axis. I also turned off all filters (high-pass and low-pass) on both the microphones and the mixer, and I tested them bare, without foam or fur (also not realistic). I recorded my tests with all mics recording simultaneously in a Zaxcom Nova2. A special note for the SuperCMIT: I recorded both the “unprocessed” channel, and the “processed” channel with “preset 1”, which is the lower-strength noise reduction recommended for regular use. I did not test “preset 2”, which is much more intense, and which Schoeps warns may contain audible artifacts (I couldn’t record both simultaneously, so I picked the one more likely to be more commonly used).

My basic audio source was my own speaking voice, recorded on-axis, and at 90° and 180° off-axis, all at a distance of roughly six feet (about as far as you’d want to get under ideal circumstances). I did the test in my living room (smallish, about 4 by 6 metres), and again outdoors in a quiet neighbourhood. In both locations, I repeated the test with an off-axis noise source (a speaker playing music) as a way to gauge how effectively the mics rejected noise. In the outdoor location, I was able to opportunistically capture some common problems: A siren, some crows, several car passes, and an airplane.

I did my best to align the recording levels of all the mics before testing, which is harder than it should be. I used my speaking voice on-axis as the test signal, and then confirmed with a 1kHz tone played from my laptop speakers. I think I got pretty close (let’s say within a dB) except for the CMIT indoors, which was 2-3dB too hot; I corrected this when I moved outdoors. It’s not super repeatable; even the 1kHz tone fluttered the levels up and down by a couple dB, but the recordings are pretty consistent when switching between tracks.

Recordings

The recordings BWF files direct from the recorder, unedited except for some file-naming. There is metadata to identify tracks, but if you are using software that can’t read it, here is the track order:

1. Schoeps SuperCMIT, unprocessed
2. Schoeps SuperCMIT, with noise reduction (preset 1)
3. Schoeps CMIT-5U ***This track was about 2-3dB too hot for the indoor comparisons.***
4. Sanken CS-3e
5. Sennheiser MKH-60

Unfortunately, I couldn’t find a suitable wordpress plugin that was capable of soloing individual channels (or even that would play BWF files), so you’ll have to download the recording and listen to them them in your favourite DAW.

Indoor Results

First, some general listening notes on how the mics compared on-axis. The unprocessed SuperCMIT and the CS-3e both sounded the most neutral to my ear, at least as far as direct sound was concerned. Tonally, the CMIT and the MKH60 sounded identical; I couldn’t tell them apart when I switched between them. To me, they sounded “better”, but also slightly coloured: Both had a bit more low end, and I liked this, but I think the SuperCMIT and CS-3e are probably more accurate. That’s my subjective opinion, so take it with a grain of salt, and you can measure the recordings if you want more precise analysis. The processed channel of the SuperCMIT sounded distinctly thin to me. And, both channels of the SuperCMIT sounded noticeably phasey or indistinct; all the other mics sounded much clearer.

I could tell immediately that, despite having the same capsule and interference tube, the CMIT and SuperCMIT’s “unprocessed channel” do not match well; they sound quite different. They are different tonally, they have different noise floors, but the biggest difference is the metallic phasiness of the SuperCMIT. I hypothesize that this comes from the additional open ports for the noise-cancellation capsule which are behind the primary capsule, and which likely change the back-pressure on the main capsule and allow an additional route for audio to reach the capsule other than through the interference tube. The noise floor on the unprocessed channel of the SuperCMIT is noticeably higher than the regular CMIT, and the processed channel is higher still (probably due to noise-cancellation capsule being in the signal chain). I hypothesize that the built-in pre-amp inside the SuperCMIT is of lesser quality than my Nova2; this can’t be changed, since the pre-amp and conversion to digital are inherent parts of how the SuperCMIT works.

On the topic of self-noise, the CS-3e was also quite noisy. Perhaps this is because, like the SuperCMIT, it has multiple capsules? The regular CMIT’s noise floor was audible, but very quiet, and I couldn’t hear the MKH-60’s noise floor at all (I’ve heard it when recording in the arctic, but rarely anywhere else).

None of the mics showed any difference in the level of the ambience in the room. I deliberately left the room quite noisy; I have a running computer right next to the mics, and the fridge is on in the background. The ambience was within a decibel on all the recordings except the CMIT, but that is because my level alignment in the CMIT was poor; the speech levels were commensurately higher than the rest of the mics, and when that is corrected for, the difference in ambience disappears. Even the noise-cancelling SuperCMIT and CS-3e did not seem to cancel any room noise, though both mics seem to emphasize a pure tone in the room noise that isn’t as noticeable in the other mics. My hypothesis is that HVAC noise is largely a diffuse noise source once it bounces around in a room, so the noise cancellation isn’t effective. Long story short, don’ t use interference tubes indoors, they don’t help. But you knew that already…

Off-axis, the mics showed bigger differences. Schoeps is known for having excellent, uncoloured off-axis performance, and that was certainly true of the CMIT. It loses a bit of high end, but unless you are listening for it, it sounds remarkably similar to the on-axis sound, just quieter. The unprocessed SuperCMIT was similar (in general, aside from the differences already stated above, the unprocessed SuperCMIT did behave similarly to the CMIT in most respects). The processed SuperCMIT sounded even more neutral off-axis than the CMIT. Off-axis, it was quieter but very close to the on-axis sound. On the other hand, it sounded thin both on- and off-axis, so even though it was more consistent, I preferred the sound of the regular CMIT.

Neither the CS-3e and MKH-60 could match the off-axis neutrality of the Schoeps mics, though their errors tended in opposite directions. The MKH-60 is a traditional interference-tube shotgun, and it sounds like it. Its directionality is primarily at higher frequencies; off-axis it sounds muffled and dark. Despite that, I was surprised how audible off-axis speech was, even at 180° off-axis. You can clearly understand what I’m saying; you could use it if you had to. With music playing in the background, this changed a bit; the MKH-60 was the least intelligible against the background noise. The CS-3e was the opposite: It’s off-axis sounds tinny, emphasizing treble and losing a lot of bass. Presumably this due to its line-array cancellation; it certainly doesn’t behave like a conventional shotgun mic. It also sounds distinctly coloured in more ways than just emphasizing treble. Off-axis, it sounded the least natural of all the mics. However, with the exception of the noise-cancelled SuperCMIT, it maintained the best intelligibility for the on-axis speech while music was playing off-axis.

One thing that surprised me was that there was almost no difference in the amount of rejection between the microphones. The processed SuperCMIT rejected slightly more noise than the others, but only by a decibel or two. In my opinion, this isn’t enough to matter, especially taking into account how much less natural the on-axis audio was. The noise-cancellation of the CS-3e made no difference whatsoever to the levels, though, as noted, it did help with intelligibility.

Overall, I felt the CMIT had the best overall balance between sounding natural and maintaining intelligibility when the background noise was competing with the on-axis speech. But, the differences between all of the mics were small. So small, that I realized I had made a mistake testing them indoors, and I needed to test them outdoors if I wanted to understand their differences. In reality, none of these mics are the right choice for indoor use.

Outdoor Results

For on-axis sound, the outdoor location reduced the differences I had heard indoors. All the same comments apply: The SuperCMIT and CS-3e sounded most neutral; the CMIT and MKH-60 sounded similar to each other, but with slightly more bass than the others, and the SuperCMIT seemed to have a slight echo. This echo was much reduced compared to the indoor test, and it seemed inconsistent; sometimes it disappeared entirely. My guess is that it is sensitive to near-field reflections, so I think it’s at its best with no hard surfaces around. Overall, whatever differences existed in on-axis sound were barely noticeable in outdoor conditions.

I made one lucky recording that did highlight some differences: While I was setting up, there was a flock of crows in the distance, cawing and making a racket. They were roughly on-axis, but very distant — probably half a kilometre away. The different ways the mics captured the reverb tail from the crows made a really big difference in how I perceived the sound. At one end of the spectrum, the SuperCMIT made the crows sound much nearer than they were: The noise cancellation was quite effective at getting rid of the ambient wind and traffic noise, as well as the echoes that helped me identify where I was. It produced a very dry recording with minimal background noise that would be perfect for a sound library. The CS-3e was not effective at removing the backgound noise, but it did get rid of the reverb tail that illustrated the sound of the space. The unprocessed SuperCMIT sounded fairly similar to the CS-3e. The CMIT and the MKH-60, on the other hand, preserved the reverb tail and the sense of space and distance. I could tell I was outdoors, and I could tell that the crows were far away. The MKH-60 was better than the CMIT in this regard: It produced the most realistic recording, meaning it was the closest to what I heard when I took off my headphones and listened to reality. Its low noise floor also helped the sense of realism.

The crows helped me identify a property of sound that is really hard to describe or quantify: I used the word realistic just now; transparent is another word. What I’m getting at is how faithfully the mic is able to reproduce the space that it is in. That comes down to the detail and clarity in the recording. I identified that realism with the ability to hear the reverb tail of the crow calls. On this basis, I was able to rank the mics from most to least realistic: MKH-60, CMIT, SuperCMIT (unprocessed), CS-3e, SuperCMIT (processed). Or, with a different recording priority, I could reverse the ranking according to which created the best recordings for library effects (best being dry, with minimal background).

I had a couple opportunities to test some common sound issues: A siren passing by in the distance, and a plane overhead. The noise-cancelled SuperCMIT was mildly useful at reducing both of these noises. The CS-3e was even milder. They were very slightly less intrusive underneath the on-axis speech, but overall I wouldn’t say they made a big enough difference to care. If I was recording in those circumstances, I would have noted the disruption at roughly the same point with all the mics. The SuperCMIT bought at most a second or two. That’s not nothing, but it’s not worth the sacrifice in realism compared to the regular MKH-60 in my opinion.

As with the indoor tests, the biggest differences showed up when comparing off-axis behaviour. I’ll go one by one, since every mic showed markedly different characteristics. As before, I did two tests: One walking around the mic well speaking as a way to judge off-axis the frequency response, and one with music playing at 150° off-axis while I was speaking on-axis, as a way of judging rejection.

The noise-reduced SuperCMIT stood out as having the most natural off-axis sound, especially beyond 90° off-axis. It tied for second best for the amount of rejection, but the sound was clear; at 180°, I could understand what I was saying perfectly when there was no competing audio on-axis. It also reduced the level of the ambience, so my on-axis voice did benefit from the off-axis rejection of the music. If I needed to hear a voice over the surroundings, but I needed the surroundings to be audible but quieter under the main subject, this was the clear winner.

The CS-3e had by far the strongest noise rejection at 180°, bringing my speaking voice down to the level of the background noise. My voice sounded thin from this angle, and was difficult to distinguish from the ambient noise. This is significant difference from the SuperCMIT, because the CS-3e did not reduce the overall ambience in the same way, so it did a better job of burying the off-axis sound in the ambience. The off-axis frequency response is terrible, the off-axis music sounded thin, unbalanced and phasey. Most of the rejection seemed to be in the low frequencies. But it did provide the best separation from the on-axis sound.

The CMIT maintained it’s reputation for rolling off neutrally off-axis up to about 90° off axis, but it performed quite badly beyond 90°. It rejected less noise than all the other mics, and the off-axis music was very muddy and muffled, to the point of being incomprehensible. The off-axis colouration didn’t sound as ugly as the CS-3e, but it was made worse by the fact that it was so much louder than the other mics. This was my least favourite choice for trying to isolate the on-axis sound in a live environment.

The unprocessed SuperCMIT gets special mention for being almost as clear off-axis as the processed version, but with the same poor rejection as the regular CMIT. It was interesting how much of a hybrid between the two it was. It was never a mix of the two mics, each characteristic took after either one or the other. Unfortunately, the lack of good rejection makes it a poor choice for booming in a crowd, but at least it doesn’t make the crowd sound unnatural.

The MKH-60 had the worst rejection at 180°, but was second-best to the processed SuperCMIT for sounding natural in that position. As with the indoor tests, it sounds a bit muffled when off-axis, but unlike the CMIT, the muffling is gradual and consistent all around the mic. It has a narrower sweet spot than the CMIT, but beyond 90°, it sounded better overall. Speech was still comprehensible; it was second only to the SuperCMIT for sounding natural off-axis. The MKH-60 had the most noticeable rear lobe, so it’s rejection at 180° was not good, but at 150° its isolation was on par with the SuperCMIT. This was the only mic where I really noticed the unevenness of the rejection pattern. I would have expected the CMIT to share this rear lobe, but, perhaps its rear rejection was so poor overall that I didn’t notice.

I have one final comparison that ties everything together: The sound of a car passing by while someone is speaking on-axis. This happened a couple times organically, and I also staged a test by driving by while music was playing on-axis. The test started with me starting my car in the background (at about 150° off-axis, 20 metres away), and it was the startup that illustrated the biggest differences in my opinion. Not surprisingly, the processed SuperCMIT had the best noise rejection, and the lowest ambience overall. However, because it wasn’t masked by the ambience, the engine startup was still audible and distracting, even though it was fairly quiet in the background. For the same reason, the road noise stayed audible longer than it might have with a higher ambience level. The CS-3e was even worse. Despite its ability to cancel noise, its off-axis frequency response made the car’s engine more noticeable and distracting. Out of all the mics, the engine startup and drive by were most clearly identifiable and distinct on the SC-3e. The remaining three samples, CMIT, unprocessed SuperCMIT, and MKH-60 were all fairly similar, and all preferable to the two noise-cancelling mics. They de-emphasized the tonality of the engine, and most of what could be heard was broadband road noise that was less distracting than hearing the engine clearly. Of the three, I thought the MKH-60 was the most balanced. I could hear a touch more tonality in the engine, but the road noise faded into the background more quickly. Even though it doesn’t have the best rejection or the best off-axis tonality, it did the best job of hiding the off-axis sound in the background, which is ultimately what I want from a shotgun mic.

Analysis

These listening tests made me re-think a lot of what I knew about what makes a good shotgun boom mic. I’ve always assumed it was pretty simple: The more off-axis rejection the better, and the off-axis roll-off should be as neutral-sounding as possible. What I’ve realized is that these are just two of many competing priorities, and favouring just those two compromises some of the other priorities.

Let me spell out some of these priorities: Isolation (the amount of off-axis rejection, measured in dB), focus (the perceptual ability to distinguish the on-axis audio from the surroundings), background suppression (how effectively the mic rejects noise from diffuse sources, i.e. ambience or room tone), reverb suppression (how dry the on-axis sound is), ability to reject a point source, realism (how natural the audio sounds), fidelity (how closely the microphone reproduces what the ear actually hears), and neutrality (frequency response, or the degree to which the microphone colours the sound). That’s a long list, so bear with me as I try to explain how they relate and why they matter.

Broadly speaking, I think there are two different approaches to recording dialogue, which I will call drama vs. documentary. For drama, the goal is to get the absolute cleanest tracks under the assumption that the recordings are just small pieces in a much larger sound design. They need have the best signal-to-noise ratio possible (isolation and background suppression), an uncoloured frequency response (neutrality), and as little reverb as possible (reverb suppression). For documentary, the goal is to get tracks that match what the camera sees (fidelity) under the assumption that the track may end up playing in the final mix more-or-less as recorded (realism), without compromising the ability to follow the action (focus). These aren’t necessarily cut-and-dried; not every drama needs to be heavily sound designed, and not every documentary needs to be verité. But they are creative choices that we make as recordists, ideally while keeping a particular post workflow in mind.

I can place the four microphones on a fairly linear spectrum between these two approaches. This is the same ranking that I discovered when listening to the crows: SuperCMIT, CS-3e, CMIT, and MKH-60. I’ll sum up my thoughts in that order.

SuperCMIT

The noise cancellation is real and effective. And it isn’t nearly as desirable as I thought it would be, because it compromises realism so much. When I purchased the SuperCMIT, I assumed (with some help from Schoeps’ marketing) that the unprocessed channel would be a close substitute for a regular CMIT. My hope was that I would get the benefits of a regular CMIT, and the bonus of noise-cancellation when I needed it. This is absolutely not the case; despite having an identical capsule and interference tube, the unprocessed SuperCMIT is a weird mix between the CMIT and the noise-cancelled channel. And, because of the phase-shifts inherent in the noise-cancellation, the two channels can’t even be mixed. I can’t see too many scenarios where I would deliberately choose to use the unprocessed SuperCMIT, but it’s a nice safety against too much noise-cancellation.

So, I had to reconsider what I wanted from the mic. Noise-cancellation is the reason this mic exists, and it is absolutely unique in that regard. The SuperCMIT stood out for the background suppression it offered: It rejects diffuse noise as well as off-axis noise, and that means it appreciably lowers the signal-to-ambience ratio. I say signal-to-ambience instead of signal-to-noise because not infrequently it removes so much ambience that only the SuperCMIT’s self-noise is left. Reading anecdotally, I found a number of complaints that the SuperCMIT is quite a noisy mic. Schoeps’ published specs barely support this; they give three different noise specs, which are only 1-3dBA above the same specs in the CMIT. This should be barely noticeable, but perceptually it seems much noisier because the noise cancellation pushes the ambience and off-axis sounds much closer to the noise floor.

One caveat: The background suppression only seemed to work in open spaces; it did not reduce background noise indoors. Even worse, indoor use created a strange double-image that made the on-axis sound phasey and indistinct. I don’t think I would want to use the SuperCMIT indoors under any circumstances other than perhaps a very large space (an open studio or an arena perhaps).

The noise-cancellation has a couple other effects as well: It suppresses reverb along with everything else, which makes it terrible for reproducing a sense of space (realism), but very good at creating a dry recording that is perfect for sound effects and post-processing. It brings the on-axis sound perceptually closer (because the relative level of direct and indirect sound is part of how we perceive distance). Put another way, the SuperCMIT has more “reach” than just about any mic I’ve heard, at least when used in an open space. I was able to zoom in on sounds in a way that other mics aren’t capable of.

Despite its very effective isolation, the SuperCMIT is only so-so for creating focus. Why? Because there’s more to creating a perceptual separation between wanted and unwanted sound than making the wanted sound louder and the unwanted sound softer. Perceptual focus is more about making the unwanted sounds blend into the background than about removing them entirely. This is similar to why the SuperCMIT sounds noisy; it takes away so much background noise that whatever is left draws attention to itself, whether it is a car starting in the background or the self-noise of the microphone. The SuperCMIT’s extremely neutral off-axis response works against it here. Frequency is important to creating a background to blend into: background noise is a low frequency cacophony that our brains tune out while we focus on the sharp, percussive, higher frequency sounds that capture our attention. By maintaining the fidelity and frequency-balance of its off-axis sounds, the SuperCMIT robs us of the frequency cues that we need to direct our attention.

The SuperCMIT is perfect for creating dry, isolated dialogue in a controlled outdoor environment. It excels at removing unwanted ambience and bringing the subject perceptually closer to the mic. I would also use it for outdoor interviews where I know the recording will play under visuals that don’t match the surroundings. The isolation comes at a price: It removes the sense of space, and makes voices seem disembodied and unnatural. It sounds accurate (in terms of frequency response), but it doesn’t sound good. I would not use it in any sort of verité environment, or on any shoot where I think there’s a chance the raw recording would find its way into the final mix without post-processing and sweetening. I wouldn’t use it as an everyday mic, but it is such a unique tool that I can see myself holding on to it for the circumstances that call for it.

CS-3e

I was surprised how much I disliked the CS-3e. In my area, this is the outdoor microphone for most mixers, on account of its reputation for superior isolation due to its ability to cancel noise via a line array of capsules. I don’t feel this reputation is warranted.

Like the SuperCMIT, the CS-3e’s cancellation does suppress quite a bit of reverb, which makes the on-axis sound drier, and helps separate the on-axis sound from the background. This gives the microphone more “reach”. Also like the SuperCMIT, this only works outdoors; indoors it just sounds indistinct and phasey. Unlike the SuperCMIT, the CS-3e didn’t seem to help with isolation very much; though it did filter out the low frequency rumble of distant traffic. The off-axis rejection is quite unbalanced, with fairly strong low-frequency rejection, leaving a very coloured, unrealistic sound behind.

This colouration is the reason I don’t feel the CS-3e lives up to its reputation. In objective terms, I believe it may measure well (reducing low frequencies off-axis probably reduces a lot of the sound power, which would be noticeable on a VU or peak meter), but it does not create good perceptual focus. By removing the low frequencies, it leaves behind an unbalanced midrange that draws attention to off-axis noise instead of hiding it. This is similar to the SuperCMIT’s noise floor problem: By removing too much, it reveals sounds that would otherwise be hidden. When I did the drive-by test, the engine starting in the background was by far the most distracting on the CS-3e because I hear it most clearly. And, not only was it distracting, it also didn’t sound good because the off-axis colouration was so severe. In absolute terms it may have had slightly less traffic rumble (I would guess at most 3dB) compared to the non-noise cancelling CMIT and MKH-60, but the price of this reduction was that non-ambient off-axis noises were much more noticeable.

Even though the on-axis performance sounded fairly flat and neutral, the amount of reverb suppression combined with the very unrealistic sound of the surroundings gave an overall impression that was quite unrealistic. In my previous experiences booming with the CS-3e for other mixers, I always had the impression that the mic sounded “harsh”. I think the lack of realism that I discovered here explains why I had that impression.

Another issue I had with the CS-3e was noise floor. It was easily the noisiest of the mics I tested, with the noise floor clearly audible in ambience, even outdoors. I’m guessing this is partly because the noise of three capsules in the design is additive, but knowing why the noise is there doesn’t change the fact that I can hear it. It may also have suffered from the same effect as the SuperCMIT, where the cancelled ambience drew attention to the noise floor that wasn’t cancelled.

I think the only use-case I can think of where I might want to use the CS-3e would be in a controlled environment (i.e. drama with a lockup) with a loud, immobile noise source (generator?), while also knowing that side noises were unlikely (and therefore the colouration wouldn’t be an issue). The 180° rejection of the CS-3e was very, very effective, much more so than any of the other mics, including the SuperCMIT, so perhaps I would use it if I needed the absolute strongest rear null I could find.

CMIT

The CMIT lived up to its reputation for having a smooth roll-off in its off-axis response. It sounded clear and neutral on-axis, and audio that wasn’t quite on-axis also sounded clear and neutral, though quieter than the no-axis sound. This was true up to about 90°, beyond which the smooth response fell apart. Sound from behind the mic sounded fairly distorted and unnatural. In practice, I don’t think this matters much, since in a typical booming position, the rear of the mic is facing the sky where there are no noise sources (other than planes), and therefore no reason to worry about it sounding unrealistic. I might choose a different mic (SuperCMIT or MKH-60) if I knew I had a sound from the rear that I wanted to reject in the background, but this isn’t a super common scenario.

Listening to the overall impression created when recording an on-axis voice in a live environment, the CMIT sounded neutral and realistic. It did a good job of separating the main subject from the background, and the background was quieter, but still part of the scene. This pretty much matches my preconceptions for what I thought a good shotgun mic should be when I started the test. It has good isolation, and it is transparent and realistic in most circumstances. I can see why it has the reputation it does, and why it is a workhorse for so many mixers. It sounded decent, but not amazing indoors, with a noticeable bump in the lower frequencies that I attribute to the proximity effect being amplified by the reverb in the room. It sounded flatter and more neutral outdoors. I could live with it indoors if I had to, but there are definitely better options (it’s a shotgun, so that is expected).

I have two negative things to say about it. One, its isolation was about 3dB poorer than all the other mics in the test. Meaning: The outdoor ambience was about 3dB louder compared to the primary signal. This had a small effect on perceptual focus, but wasn’t a deal-breaker in my opinion. Perceptually, I found the noise-cancelling mics to be worse for focus because of the way they cancelled the background ambience but left behind off-axis noises (even if those noises would have measured quieter than the CMIT). Two: The noise floor was audible in the ambience. I wouldn’t say this was a major issue; it was quieter than both the CS-3e and the SuperCMIT. But it was there. My test was in a fairly quiet outdoor neighbourhood, so there are probably many places I could have recorded where the noise floor would have been completely buried.

Overall, I would have no hesitation using the CMIT as my default outdoor boom mic. It sounds great, provides isolation, and is forgiving off-axis. I especially like it for documentary scenarios where I can’t always anticipate who is talking next; the smooth off-axis and lower isolation is actually helpful in this scenario, since it is more forgiving if I don’t anticipate who needs to be on axis. However, I probably wouldn’t choose it in an uncontrolled environment (e.g. a crowd) where separating the voice from the environment is the primary need.

MKH-60

The MKH-60 stood out to me as the best overall compromise between the various priorities that I identified in my preliminary analysis. It didn’t stand out as the best for any individual trait except noise floor, but I think it is the mic I would choose first for the broadest number of scenarios.

I thought it was the most transparent, realistic-sounding mic. This was most apparent to me when listening to the crows: Their calls had the longest, nicest-sounding reverb tail, which also happened to be the closest to what it sounded like when I took my headphones off and listened to the live sound. It gave me the most realistic sense of space of all the microphones. In most scenarios, this was a pretty impression. In many respects, it sounded very similar to the CMIT, and I had trouble distinguishing the two mics at times (usually when there was no obvious off-axis sound to compare). The extremely low noise floor also helped with transparency; it is the only mic whose noise floor I couldn’t hear in any of the tests. In fact, the only time I have ever noticed its noise floor was when I was recording in the arctic, 400 kilometres from the nearest road.

It is not as smooth off-axis as the CMIT; it sounds dull and slightly muffled from pretty much every angle off-axis. On the plus side, once the sound is off-axis, it sounds consistent and doesn’t change much from front to back the way the CMIT does. Sources from the rear sound much the same as from the side or as little as 30° off-axis. This is the classic shotgun sound: good high-frequency isolation but not much low-frequency directivity.

I didn’t realize until this test how useful this characteristic actually is. Even though it’s objectively less neutral, the lack of low frequency isolation actually helps the shotgun do its primary job, which is to create perceptual focus on the on-axis sound. The high-frequency roll-off means the off-axis sound all mushes together into a background rumble, and it becomes much harder to distinguish individual sounds compared to the on-axis sound. It prevents off-axis sounds from being distracting because our brains tune out the muddy rumble and focus on the mids-and-highs that stand out from all the rest. I have changed my opinion about wanting my shotgun mics to have neutral off-axis frequency response. This would be desirable when recording music or ambience, but it is not desirable in a mic that is intended to direct the listener’s attention.

This conclusion comes directly from what I experienced in the outdoor listening tests, especially the car drive-by. In this test, the engine starting in the background was the least distinct of all the microphones: I could barely hear it, and I wouldn’t have noticed it at all if I wasn’t looking for it. This is very different from the CS-3e, where the engine start-up was crystal clear and quite distracting. I daresay the background ambience in the MKH-60 was objectively louder, but it was easier to tune out and less distracting.

It also sounded more natural. I would say the CMIT, and especially the SuperCMIT had better fidelity (i.e. they were closer to what the car startup sounded like in person), but the MKH-60 was closer to what I expected to hear when I was trying to listen to the dialogue and not the overall soundscape (which is the normal scenario for a shotgun recording).

The muffled off-axis response does have a disadvantage. My ability to comprehend speech was slightly worse than with the other mics when competing with background noise. I attribute this to a combination of less reverb suppression and the background ambience being more broadband. In a very noisy or heavily reverberant environment where every dB of separation counts, it might make sense to sacrifice the realism of the MKH-60 for the better isolation of the noise-cancelling SuperCMIT. But it wasn’t a big difference, and in most scenarios I would pick the better realism and perceptual focus over pure comprehension.

Conclusions

When I started this test, I was not expecting that my workhorse MKH-60 would end up being my favourite. This is the cheapest, most unassuming and “normal” shotgun of the bunch, and it’s out of production to boot. Most of the other mixers I know use the more expensive CMIT and CS-3e as their default shotgun mic, and I assumed that there was good reason for that preference. I have learned to disagree. Knowing what I know now, I will continue to use the MKH-60 as my everyday shotgun mic. It is the cleanest, most detailed, most realistic sounding mic of the bunch, and I really liked its ability to focus attention.

I’m having trouble deciding whether I want to keep the CMIT. The CMIT and the MKH-60 are quite similar; the differences are subtle, and there aren’t a lot of scenarios where one isn’t as good as the other. But difference do exist, and, although I prefer the realism of the MKH-60, I think the more forgiving off-axis performance of the CMIT might be useful considering the amount of unscripted documentary I do. While I’ve never had any complaints about the MKH-60 doing documentary, I think the CMIT is a slightly better choice for run & gun situations where I’m following multiple characters and I don’t anticipate needing to reject a lot of off-axis sound.

I did not anticipate how much I would dislike the CS-3e. To my ears, it just doesn’t sound good, and its the only mic I can’t see keeping in my kit for special circumstances. I confirmed my subjective impression that it sounds harsh, and, despite the noise-cancelling design, I found that its off-axis performance made it more distracting, not less. And, for anything where I really do need more reach or noise-cancellation, the SuperCMIT was both more isolated and less coloured.

I don’t love the SuperCMIT as an everyday mic. In most scenarios, I don’t like how artificial it sounds, and the artificiality is a direct consequence of its noise-cancelling design. I find the lack of “space” disorienting. But, it is doing exactly what it is designed to do, and there are scenarios where a very dry, clean signal is exactly what is needed. It is exceptional at separating the on-axis subject from the background noise, and it has more “reach” than any other mic I’ve experienced. I would use it for sound effects, and in noisy environments where I know post will be processing it heavily anyway. It’s good at suppressing reverb in open environments (though not most indoor locations), and I can see myself using it in spaces like gyms, arenas, or churches. I might also use it for outdoor “wilderness” interviews when there is traffic nearby, or when I know it will be edited under visuals that aren’t outdoors. There isn’t really anything else like it, so I would keep it for the scenarios where I know it can do something that other mics can’t.

I was surprised how little difference there was in isolation or the ability to suppress background ambience between the mics. The SuperCMIT was slightly better at isolation than the others, and the CMIT was slightly worse, but I was expecting the amount of isolation to be a major source of difference, and it wasn’t. Moreover, I discovered how unimportant the amount of isolation is perceptually. Making the background a couple dB louder or softer had very little effect on how I perceived the recordings; for the most part, my brain just tuned out the background and focussed on the foreground. There is certainly a threshold where the background would be come too loud and would start to affect comprehension, but for the most part, good recording technique would be way more important than the amount of isolation. Yes, there are sometimes technical reasons when every dB counts, but for the most part, I found that the frequency balance of the ambience and off-axis sound played a much larger role in how easy it was for my brain to focus on the main dialogue.

I was also surprised how little difference there was in on-axis response. If I wasn’t listening to the background, or there wasn’t much in the background to listen to, all of the recording sounded very similar, and frequency response simply wasn’t an issue. For the amount of attention it gets on the spec sheet, frequency response simply wasn’t a factor in differentiating the mics. Perhaps that’s because I’m testing a slate of very high-end mics, but I’ll give it less weight when thinking about mics in the future (and, it’s usually easy to change with EQ anyway).

Finally, although it’s common sense (at least among professional mixers) that shotguns aren’t intended to be used indoors, I somehow assumed that I could make useful judgments between them by testing them indoors. This was a mistake, and I mostly didn’t consider the indoor performance in my analysis and conclusions. But, I leave the tests as an example of how ineffective they were. It was also useful to hear just how bad the two noise-cancelling mics sounded indoors; both sounded robotic and phasey trying to deal with room reflections. I was a bit surprised that neither noise cancelling mic had any effect on the level of the room tone; my tentative conclusion is that room tone is so diffuse and omnidirectional that the noise-cancellation simply couldn’t operate.

These tests challenged what I thought I knew about what a “good” shotgun mic is, and they have changed how I think about my goals for “good” sound. I’ve never had the luxury of a large mic cabinet before, but I now have a much more nuanced idea of which priorities I’m trying to balance when I choose a mic. Do I need the most isolation (probably not). Do I need to suppress background ambience, or a specific off-axis noise? How dry do I need the subject to sound? How much do I want to place the subject in the space that I’m recording in, and how much do I want to abstract the subject from its surroundings so it can be processed in post. These are all questions I thought I knew the answer to before, and which I now have different answers to. I hope that my learning with be your gain as well.