About the size of a large guitar pedal, the Decoder is an all-analog box that lets you convert mic or line-level Mid- Side signals to stereo Left/Right. It's relatively easy to set up a digital matrix to convert between M-S and L/R in your DAW, and you can also set up a matrix using three channels of a mixer, but the Decoder lets you do this in the analog domain, pre-conversion, with greater ease (and more precision) than mult'ing and polarity-flipping through an analog mixer.
Why would you use an M-S arrangement? The main reason is that M-S, unlike any other stereo technique, has the distinct advantage of being 100% mono-compatible. After the M-S signal is converted to L/R stereo, the differences between the L and R channels will cancel to zero if the L/R signal is summed to mono, leaving only the original mid signal as if you used a single mono mic.
I find that micro-differences in mic placement can really affect the stereo field, so stereo tracks converted from M-S can sometimes "pull" at my ears, like out-of-polarity speakers do. If you get the mic placement right though, you're rewarded with a very cohesive, involving stereo image that's usually more defined than an X-Y recording of the same situation. An adjunct benefit is that if you record the mid and side signals directly to your recorder, you can decode them to standard stereo after the fact, thus giving yourself control over the amount of stereo information later.
The Decoder accepts mic or line-level signals, so not only can you insert its M-S matrix into your recording chain, but you can also utilize it later during mixing. Plus, if you route the sends from your record channels to the Decoder while you record in M-S, you can monitor in decoded L/R stereo, for the best of both worlds.
The box itself is made with heavy-gauge steel, using Radial Engineering's signature "book-end" chassis design. Most of the controls are protected between the protruding edges of the wrap-around "hardcover." There are no meters or clip lights. Some switches, including the pushbuttons for phantom power, are recessed and therefore require an implement such as a tweaker or paperclip to operate. This got tiresome when auditioning mics and forgetting where I just put the tweaker! (Radial's reasoning is that you won't damage your ribbon mic or source device by accidentally turning on phantom power, and there's less chance you'll blow up your speakers by accidentally switching from line to mic-level.) While the unit looks like it could literally be run over by a truck without affecting the settings, the knobs do flex inward, which means they are not affixed to the metal case. There's enough room between the hardcover ends to kick the knobs with the toes of my shoes. I refrained from trying that on purpose, but as the unit often ends up on the floor when in use, this may be the one weakness in the otherwise military-strength construction.
Traditional M-S recording employs a single cardioid mic for the mid channel and a bidirectional figure-8 mic for the side. To use the Decoder in this manner, you plug the mid mic (or line-level) source into Input 1, and the side mic/line into Input 2a. A standard L/R stereo line-level signal then appears on the left and right outputs of the Decoder. Gain pots let you adjust the mid and side levels being fed into the matrix. In an interesting twist, the Decoder also includes an Input 2b, allowing two separate mics — for example, a pair of cardioids facing outward — to be used together to produce the side channel. A stereo pair plus a mono mic? Do you lose the benefits of M-S recording if you do this? If the two side mics are well-matched cardioids placed close enough together, this ends up being not exactly but practically the same as using a single figure-8 side mic. As a test, I connected a single lollipop-style multi-pattern condenser mic, switched to figure-8 mode, to the Decoder - without a mid mic. The resulting L/R stereo signal canceled completely when summed to mono. So then, I connected a pair of the lollipop mics, both switched to cardioid, into the two side channel inputs of the Decoder, pointing in opposite directions with their screens almost touching and their diaphragms about 1'' apart. The resulting signal did not cancel completely, but almost. What does it sound like to spread out the cardioid side mics? In my trials with acoustic guitar, not that good. The sound went from tight and cohesive to blurry and hollowed-out just by making the two side mics a spaced pair. It was not a sound I was looking for, but might work for something that doesn't need (or want) focus, like a set of room mics.
Is M-S good for anything else? Yes. You could think of M-S as "subtractive" stereophonic synthesis. As long as the mid and side signals have some overlapping frequencies, something is going to happen between them in the stereo field. So, using the Decoder, I tried things like routing two different delay lines to the mid and side. This produces an organic undulation between left and right when the repeats
overlap. If you EQ the delays differently, that changes the way the stereo field sounds and changes shape. I also tried using different reverbs for mid and side. When the reverbs were very different, it gave a "swimming" quality to the stereo field. The sky is the limit for how creative you want to get. When sending a dry vocal to the mid, and mono reverb to the side, it seemed easier to get the right balance between wet and dry, and the reverb had a stereo feel even though it was actually mono. I also tried processing a clean guitar track with a doubled overdriven track through the Decoder. With overdriven guitar in the mid input and clean in the side, it created a highly defined yet totally overdriven and very aggressive sounding guitar, that subtly shifted from side to side depending on what was played. With clean mid and dirty side, I found I could get great bluesy overdrive, and clean sounds that would also subtly shift in the stereo field. It was really fun to experiment this way with headphones on, and it's a good use of the analog, zero-latency nature of the box. I highly encourage tone- questers to give the Decoder a try.
There is one huge downside to non-traditional uses of M-S though: Whatever you do with the side signal is going to cancel out if the stereo mix is summed to mono. So, in the vocal example above with reverb assigned to the side input, the reverb drops out in mono. This is less and less of a concern as the years go by, but should be noted. (If your mixes make it to radio, you could also have some fun with mono AM radio stations getting very different results from stereo FM.) I also found that when changing configurations, I had to refer to the manual, because the Decoder is more than just a simple 2×2 matrix, and it's not 100% obvious from the panel labeling how it all works.
By the way, you can't disable the M-S matrix, so the Decoder won't work as a straight-through two-channel preamp. However, you can use it as a one, two, or three-input mono mixer, which came in handy when I mic'ed a guitar cab with three mics. (Input 2b's polarity-reverse button facilitates this kind of use.) The mic preamps themselves sound pretty good to my ears and seem to fall into the new class of low-noise, wideband, everyday preamps that I keep encountering. I think an opportunity was missed to be able to switch off the M-S processing altogether, so as to have a portable, two-channel preamp in the bargain too. I am also missing a battery compartment; if it had one, this would be a great self-contained field-recording preamp. (It takes power from a 15 volt DC wall-wart, so conceivably, you could power it with an external 12-15 volt battery pack.)
If you record to a DAW, the Decoder is more convenient and doesn't add latency compared to a matrix implemented within the DAW. If you record away from a DAW, or you patch in analog gear for mixing, the Decoder offers utility and creativity for significantly less money than the next cheapest non-DIY active M-S decoder I could find, and it's likely higher-fidelity than a non- dedicated analog solution that you would patch together. I think Radial Engineering has a winner here.