Results and Discussion

Well, now I've got to the hard bit - actually giving an opinion!

It works...

Basically, it's simple - this microphone, in combination with the software, does what it was designed to do - it produces a B-format representation of the sound field it is placed in.  That's to say, it has no particular character (character in the context of audio, usually means some form of distortion or unnaturalness which in some situations pleases the person using the word), and presents no magic properties

In essence, I can use this microphone system as a tool.  So long as I use it within its design limits it simply does what is required and virtually nothing else (this corresponds to Doug Self's concept of the blameless amplifier).  It is easy to transport and to set up, and so is ideal for the location recording that is most of what I do. Should I need, I can rotate the soundfield with no harmful effects on the sound, and I can generate not only ambisonic recordings, but stereo using different techniques (Blumlein crossed figures-of-eight, or MS, for instance; or with a little more effort, UHJ).

Frequency Response

A flat frequency response is the main thing that leads to a microphone being characterless.  Here I must admit to having some privileged inside information, as I have been allowed (under NDA) to see the resulting B-format frequency response graphs plotted during the calibration of my particular microphone.  They are not suitable for publication, because they are not straightforward "marketing" graphs; it is actually very difficult to measure microphones, especially figure-of-eight ones like the X, Y and Z components of B-format, and this is reflected in the plots.  However, what is apparent is that over the main part of the frequency range all the component outputs are extremely flat; their on-axis curves are flatter than those of my AKG C414s (which come with individual on-axis plots); the off-axis curves show distortion of the polar patterns above about 8kHz, but this also occurs with "real" microphones though the shape may be different.

LF Extension

The "velocity" aspect of directional microphones has an inherent bass fall-off.  This is usually allowed to manifest itself below about 200Hz, because typical modern close usage causes a bass rise because of the proximity effect.  However, for ambisonic recording at audience distances, this bass fall-off is not appropriate, and so in B-format the bass is arranged to be maintained down to about 25Hz.  Very few purpose-made figure-of-eight microphones can get close to doing as well - the BBC-designed Coles 4038 ribbon is one (40Hz), and the AKG C414 is another (55Hz).  The TetraMic is calibrated to maintain its bass to 25Hz in all channels of the B-format.  As I like to record organs, this matters to me!  However, I haven't yet used the TetraMic to record an organ - that opportunity comes up in a few weeks time.

It is common to suggest that we cannot use directional information at the lowest frequencies, below about 100Hz, and so it is not useful to reproduce them - hence the prevalence of single subwoofers in domestic surround systems.  However, such systems cannot reproduce, for instance, the trouser-flapping effect of low organ pipes, producing instead oppressive pressure signals only.

The only downside of a full LF extension in the directional channels is that they are made much more sensitive to disturbances such as wind, knocks and earthquakes than most microphones - this is not a flaw, but a result of the physics involved.  Hence the importance of always using a firm stand and a good shockmount.  When second and third order microphones become practical, this effect will be correspondingly worse.  VVMic includes a variable LF cut to reduce the LF in case of problems.

HF Equalisation

Because of the distortion of the B-format polar diagrams in the top audible octave that is inherent in tetrahedral soundfield microphones, it is not entirely clear what the ideal equalisation of that top octave is.  In simple microphones, a flat on-axis response leads to a diffuse-field response which reduces at high frequencies as the polar diagram narrows; the polar diagram irregularities of a tetrahedral microphone are more lumpy than those of corresponding "real" microphones, but because the tetrahedral microphone doesn't really have a back, the diffuse response is probably closer to the on-axis response.  Referring back to the measurements of my microphone, I note that the directional components have been allowed to have a peak of a couple of dB at around 14kHz in some directions, to compensate for a slight fall in other directions.

My impression is that this has been judged about right; but my ears are not as good at those frequencies as they were 40 years ago when I worked at the BBC, and I need experience in a wider range of recording venues to come to a firm conclusion. So far all my TetraMic recordings have been in medium to dry venues; my next recording will be in a college chapel, with lots and lots of ambience, and this will be a particularly interesting comparison as I have several previous recordings from the same place.


The TetraMic outputs are noisier than my AKG C414 ULS (let alone the incredibly quiet current version); they are similar to the AKG Blue Line mics I have used for native B-format recording (see elsewhere on this site for details).  In practice, this is quiet enough.  I am never conscious of microphone noise as a limiting factor, even when recording quiet piano music - background noise in the venue is almost always the limiting factor, even in a studio.


In my usage I am never likely to approach this, though one person has remarked on possible distortion on one of my recordings (of a female voice at about 1.5m).  I have yet to to follow this up in detail, but I certainly have certainly not been conscious of it in my own listening.


The VVMic software does its maths flawlessly, as far as I can tell.  There have been a few niggles - an early version could be persuaded to lock up, sometimes the output files are 15 samples shorter than the input files, and there are a few small inconsistencies in the handling of file names - but these are all trivial matters, and may have been improved in the latest version (which I haven't tested for these points yet).  In any case, I have to admit that I haven't even got around to reporting them.
Ambisonic Info header