Voyage Audio Spatial Mic

The Spatial Mic with basket removed, revealing its array of eight capacitor capsules.

This debut mic from Voyage Audio combines the power of Ambisonics with the convenience of direct USB connectivity.

Quadrophonic stereo was an expensive failure in the '70s. Music-industry money was thrown at 5.1 surround in the early 2000s, but it never caught on. Is it third time lucky for surround sound, and what's changed?

The current interest in surround formats is driven less by the music industry, and more by film, TV and gaming. One thing that's changed is that it's now practical to stream the large quantities of data involved in real time, but probably the biggest development is that distribution formats are no longer tied to a specific listening format.

Quad and 5.1 systems delivered a single channel of audio for each speaker. If you invested in the appropriate four, five or six-way speaker and amplifier setup, you could enjoy surround sound. If you didn't, or you wanted to listen on headphones, tough.

By contrast, modern formats such as Dolby Atmos and Ambisonics don't care what system you're listening on. They describe a 360-degree soundfield in a signal that can be decoded to suit any playback system, whether that be a mono phone speaker, a pair of cheap earbuds, a home stereo, a small surround rig or a full-blown cinema system.

Ambi Dextrous

I've described these formats as 'modern', but in reality, that adjective only really applies to Dolby Atmos. Ambisonics, the brainchild of British audio genius Michael Gerzon, has been around since the '70s — it's just taken the world that long to catch up with its possibilities. And, unlike Atmos, Ambisonics is a capture format as well as a delivery format, and its first practical application was in the groundbreaking Soundfield microphone.

The four-channel 'B‑format' Ambisonic signal from a Soundfield mic can be decoded to a surround speaker array, but it can alternatively be decoded to one or more 'virtual microphones'. This means that you can record the output of a Soundfield mic and decode it later to recreate any coincident mono, stereo or surround mic array, pointing in any direction. In other words, an Ambisonic microphone is an incredibly useful tool for music recording, even if the results are only ever going to be released in stereo!

All of which brings me to the subject of this review: the Voyage Audio Spatial Mic. This, too, is an Ambisonic microphone, but whereas the Soundfield mic has four capsules and outputs a four-channel signal, the Spatial Mic doubles this count. It is, in other words, a 'second-order' Ambisonic microphone.

To offer a crude analogy, the move to higher-order Ambisonics is the sonic equivalent of moving up from DVD to 4K TV, at least in theory. A second-order Ambisonics signal describes the 3D soundfield in more detail than does a first-order signal, allowing the positions of individual sources to be pinpointed more accurately. On paper, at least, that should make it possible for the listener to discriminate sources that would blur together in a mere first-order recording; or, alternatively, to achieve the same level of separation from a greater distance. On the downside, a second-order Ambisonics recording presents twice as much data to handle as a first-order one, and means that you may be dealing with the cumulative self-noise of eight mic capsules rather than four — although this is not a straightforward calculation.

A first-order Ambisonic recording can be thought of as an extended Mid‑Sides signal, containing the hypothetical output of three figure-8 microphones arranged along the X, Y and Z axes, plus that of an omnidirectional mic. It's not practical to actually position three figure-8 capsules coincidentally, though, so most 'Ambisonic' mics, including the Soundfield, synthesize these signals from four subcardioid capsules arranged tetrahedrally.

The analogy with Mid‑Sides recording breaks down with higher-order Ambisonics, because the information carried by each of the additional channels no longer relates to a conventional microphone polar pattern. What remains true is that the Ambisonic signal is derived from conventional directional mic capsules, mounted as close together as possible.

The Digital Trap

The Spatial Mic is a smart and surprisingly lightweight device which has the look and feel I typically associate with USB microphones. Which shouldn't surprise, because the Spatial Mic is a USB microphone. A Type C port on the base of the mic allows it to be connected directly to a computer, in which case it is bus powered and appears as a class-compliant device presenting an eight-channel signal. Alternatively, it can supply those eight channels over an ADAT optical connection, but you'll also need to feed it 5V through a separate Micro USB connector. ADAT mode is 48kHz only, whereas the Spatial Mic can operate at 44.1, 48 or 96 kHz over USB.

In USB mode, the Spatial Mic also delivers stereo audio from your computer to a mini-jack headphone socket. This computer audio is blended at the level of your choice with a fixed binaural decode from the mic itself. This blend, headphone volume and the gain of the mic's internal preamp circuits are set using the chunky press-and-turn encoder with its large LED surrounds, which also serve for metering purposes. This is all simple, quick and effective, with the downside being that the mic's metal shell is quite resonant, and even the lightest touch is picked up quite loudly. No shockmount is provided, just a simple standmount that screws to the reverse side of the shell.

The mic's output arrangements are probably the biggest factor limiting its usefulness for music recording. Unless you want to get into creating Aggregate Devices in Mac OS, connecting over USB limits you to using the Spatial Mic as your only audio interface, meaning that you can't record it alongside other mics on a session. If you want to do that, ADAT is your only option, and the logistics of positioning the mic in such a way that it can simultaneously receive 5V power and feed an ADAT socket on your interface can be challenging. Since it has no ADAT or other clock input, the mic also has to serve as your system's master clock when used over ADAT.

Once captured, a recording made with the Spatial Mic becomes an endlessly fascinating sonic playground...

From A To B

As with the Soundfield, the audio that emerges from the Spatial Mic simply consists of one channel for each of the mic's capsules. This 'A-format' signal is not an Ambisonic recording: before it can be treated as such, it needs to be matrixed into a 'B‑format' signal. The mic itself puts out eight channels, but second-order Ambisonics requires nine, and matrixing is handled by a free plug‑in called Spatial Mic Converter. Initially, this was a rather incomplete solution, but a comprehensive update shortly before this review was completed added many new features, turning Spatial Mic Converter into a one-stop decoding shop that's as easy to use as the equivalent Soundfield By Rode plug‑in.

In essence, there are two possible applications for a device like the Spatial Mic. Perhaps the most obvious one is to use it to capture 360-degree audio to accompany 360-degree video. In this role, you'd simply matrix the Ambisonic signal into the appropriate format for a delivery platform such as Facebook 360, where it can be decoded on the fly. The advantage of a second-order over a first-order Ambisonic mic in this role is better localisation, allowing the listener to identify the position of individual sources more precisely.

The other main application is to decode one or more virtual microphones, pointing in the directions of your choice, for use in a conventional stereo mix. Like the Soundfield, the Spatial Mic can replicate any coincident mono, stereo or multimic array using first-order polar patterns. Or, to put it another way, the Ambisonic signal can be used to generate any number of virtual mics, pointing in any direction, having any polar pattern on a range that runs from omni at one end to figure-8 at the other, with subcardioid, cardioid and hypercardioid as intermediate points.

What a second-order Ambisonic mic can do that a first-order one can't is go 'beyond' figure-8 on that scale to deliver second-order virtual polar patterns.

New Directions

As I mentioned earlier, the additional channels in a second-order Ambisonic recording don't neatly correspond to any real-world pickup pattern, and neither do the additional virtual mic patterns that can be derived from them. A key term in Ambisonics literature is Re, the coefficient of directionality. The more nearly this value approaches 1, the more discriminating the mic in terms of its directional capture. An Re value of 1 would be achievable only in an Ambisonics system of infinite order, while the upper limit on Re in a first-order Ambisonics system is about 0.577. Moving to second-order Ambisonics raises this theoretical limit to 0.775.

The Spatial Mic Converter plug‑in allows you to synthesize any virtual polar pattern.

The Spatial Mic Converter plug‑in operates in four modes, of which two are labelled 'Max Re' and 'In-Phase'. These both decode to the pattern that maximises Re — in effect, a kind of tail-less hypercardioid — but the second does so within the additional constraint of phase coherence across all the output channels. 'Figure-8' is the familiar first-order mode in which the virtual mic pattern runs from omni at one extreme to figure-8 at the other, with cardioid in the middle. 'Second order' mode likewise runs from omni to cardioid, but thereafter moves towards second-order cardioid rather than figure-8. At these settings, a more narrowly focused, beam- or shotgun-like unidirectional pattern is achieved, at the cost of a rather weird multi-lobed rear response.

From the perspective of music recording, an Ambisonic mic is perhaps most useful in ensemble recording, where its ability to provide a stereo capture that can be 'steered' after the fact is invaluable. Lockdown meant I wasn't able to test the Spatial Mic in that application, but I could be my own guinea pig for one of my other favourite uses for B‑format miking: recording a singing guitarist. The beauty of using a single Ambisonic mic for this purpose is that you can derive separate virtual mics for the vocal and the guitar, and move these around to maximise separation or simply get the nicest guitar sound. All the virtual mics remain coincident, so there's minimal comb filtering when they're combined at the mix or collapsed into mono.

Space Travel

Placing the Spatial Mic alongside my own Soundfield ST450 MkII and decoding to standard first-order polar patterns revealed a noticeable difference in the tone of the two mics. At a distance of 30cm or so, the Soundfield was full and rich, with plenty of low midrange; the Spatial Mic was a fair bit leaner in comparison. In some ways it was actually closer to a 'mix ready' sound, though I suspect that adding in the missing warmth on the Spatial Mic would be harder than cutting it from the Soundfield. In general, while I'd be perfectly happy to use the Spatial Mic in a mix, there's a crispness to its sound that begins to grate under some circumstances.

When I tried to extract separate vocal and guitar signals using the different types of virtual mic available, it was immediately clear that you can get more separation using the Spatial Mic than is available from a first-order mic. However, it was also clear that if this is your goal, you can pay a penalty in terms of sound quality. When you focus the Soundfield into its narrowest patterns, either by decoding to figure-8 or using the beam-forming option on Rode's Soundfield By Rode plug‑in, the off-axis spill onto the virtual mic tends to be pretty benign and subdued in tone. By contrast, although the 'pointiest' patterns in Spatial Mic Decoder give noticeably less off-axis spill, the tone of that spill is more objectionable. There are also surprisingly large differences in output level between apparently similar virtual mics decoded in the different modes. The most extreme patterns tend to give the on-axis source a slightly processed sound, and I think they're probably best reserved for applications such as news gathering where intelligibility and separation are the highest priorities.

However, that's not to say that the Spatial Mic offers nothing over a first-order Ambisonic mic in music applications, because there's a lot of ground to be covered between the limits of what a first-order mic is capable of, and the extremes where maximising directionality comes at a cost. I don't think that the availability of second-order patterns will revolutionise miking in a music-recording context, but it does add even more versatility to the already hugely flexible Ambisonic concept. Once captured, a recording made with the Spatial Mic becomes an endlessly fascinating sonic playground, and you may have to exercise some self-discipline to stop trying different decoder settings and get on with your mix! I was also pleasantly surprised by the Spatial Mic's self-noise. Subjectively, this was no higher than that of my Soundfield through its analogue B‑format matrix/preamp and an audio interface, and wouldn't be an issue in normal use.

Verdict

The Spatial Mic costs around the same as Rode's excellent NT‑SF1 first-order Ambisonic mic, and about half as much as the only other second-order Ambisonic mic I know of, Core Sound's Octomic. In both cases there are two key differences. The NT‑SF1 and the Octomic are robustly built tools intended for the rigours of professional use, and both are purely analogue devices. The Spatial Mic's less confidence-inspiring construction is forgivable at the price, and with only digital outputs on offer, it doesn't introduce the additional expense of matched preamps or A‑D converters. Unfortunately, though, its self-contained nature is also the biggest obstacle to its use for music recording. The Soundfield is almost never my only mic on a session, and in its present incarnation, that's the only easy way to use the Spatial Mic.

That's not exactly a criticism, because I can't see how else Voyage Audio could have brought it to market at such a low price, but it does make the mic less useful in real-world circumstances. How about a more solidly built 'pro' version that could be powered and connected to an audio-over-IP network using an Ethernet cable?