Modelling microphone systems are one of the really significant developments in pro audio of the last few years. The promise of being able to accurately model microphones as accurately and convincingly as guitar amps and speakers is a compelling idea but one, which until very recently, was out of reach of most engineers and producers.
As is often the way with such things, the reality of viable microphone modelling seemed to happen quickly when it finally came, with Townsend Labs and Slate delivering their systems almost at the same time in 2016 with Antelope arriving with their system about a year later. Superficially all the systems seem similar, they all offer a workflow where a system-specific reference microphone is used for capture and the timbre of a collection of classic microphones can be imposed on this recorded signal either live at the recording stage or to give the user ultimate flexibility: Being able to record once and audition a cupboard full of eye-wateringly valuable (and in some cases unavailable) mics at the mixing stage.
With such a tempting proposition on offer, the first question everyone wanted to know was whether the models sound like the original microphones? With vintage microphones there is, of course, a variability between examples of the same model of mic and small differences are to be expected.
As with modelled studio hardware, there will always be some people who will never accept anything other than the real thing, and likewise, some people will never be convinced by modelled microphones because they are modelled microphones. However, for the rest of us, there seems to be a consensus that modelled microphones do indeed sound very much like the mics they seek to model.
4 Different Microphone Modelling Systems
The four systems compared here are the Slate ML-1, the Townsend Labs Sphere L22 and the Antelope Edge Duo and Edge Solo. This doesn’t cover every modelling microphone made by these manufacturers as not every mic has been tested by Audio Test Kitchen.
To find out more about these modelling systems read our article Microphone Modelling - Antelope, Slate, Townsend Labs & More - Our Pick Plus Take Our Poll
Audio Test Kitchen
So modelled mics aim to sound like the mics they seek to model, but a more controversial question is which of the systems currently available is the most accurate? That is a question which might to some extent be answered by looking at the data from Audio Test Kitchen, an impressive repository of microphone data who have tested not only the Townsend, Slate and Antelope mic modelling systems but also some of the mics that all three manufacturers have modelled.
If you are unfamiliar with Audio Test Kitchen, it is a very clever microphone comparison resource that uses a combination of techniques to reproduce the same performance over and over again and records it using many different microphones under identical conditions. The results can be used to compare recordings made using these mics against each other on a variety of different material and instruments. Audio Test Kitchen publish detailed frequency plots for the mics they test and using these as the basis for a comparison has been very interesting. To hear the difference between the mic models, and any other mics they feature head over to Audio Test Kitchen where audio examples are available. To audition these examples sign up for a free account at Audio Test Kitchen first and you’ll be able to hear the models for yourself.
Audio Test Kitchen are unbiased, having no particular interest in any one system over the other, and they test and compare microphones all day, every day. They have a carefully designed procedure and will do a far better job of this kind of detailed, empirical testing than we could ever hope to.
Which Mic Model To Choose?
There is a lot of common ground between the models offered by the different systems available. This is no surprise as there are most definitely some superstar mics out there which everyone would like to own, however while it might be tempting to use something like a U47 as the basis for the comparison, if we are to compare the models to the performance of the actual mic on which the models are based, these vintage mics are the worst choice as there will be a significant variance between examples of these decades-old microphones. The Sony C-800G is a far better mic to look at because it is still in production and made with relatively tight tolerances. An example of this mic is far more likely to be a good basis for comparison. Luckily all these systems offer at least one Sony C-800G model.
In the plots above it is clear that there is some significant variation between the modelled mics and the Sony, shown as the trace in bold. Note the vertical scale, the biggest deviation is approximately 4dB. If we look at these in turn we can see some significant differences.
Here's the Sphere L22 LD800 frequency response compared to the actual C800G. There is a small difference in the 2k to 7k range which is probably within the manufacturer’s production tolerances and could be found between any two C-800Gs. The low frequency response is very accurate indeed.
The OW800 for the Sphere L22 is even closer at high frequencies and a little less at low frequencies, however this model is from the Ocean Way collection and is based on Allen Sides' prototype C-800G so this difference might not be incorrect. Allen Sides and Ocean Way were involved in the development of the C-800G.
The Slate ML1 FG800 model shows some top end differences between it and the Sony especially at 6kHz.
Slate also offer the FG800M model. There is still a significant difference around 6kHz. The FG800M is an alternative model, modelled on an older example and is recommended by Slate for people who “love the FG800 but wish it was a tad less sibilant and bright”.
The Antelope Edge Duo 800T has some little dips at 320Hz, 1400Hz, 4500Hz. And then the bass response deviates a fair amount below 75Hz which, given its shape could be proximity effect differences. If encountered in the real world the user would probably compensate by working a little closer to the mic.
As can be seen from the plot, the Edge Solo shows some significant deviations from the reference mic.
Is There More To The Performance Of A Mic Modelling System Than Just Its Frequency Response?
In this article, it is our intention to present the data from Audio Test Kitchen and as much as possible to let that data speak for itself. Because their data is based on real recording of real instruments Audio Test Kitchen offer audio examples so you can hear the different models as well as view frequency plots. However, while we aren’t going to make any judgements on the relative merits of the microphone modelling, there is some context it would be useful to offer.
Proximity effect comes into play and as the contribution of the proximity effect varies with distance, differences at the bottom end in this test might well be a consequence of using an identical distance in each case, in the real world a user would just move the mic slightly closer. Single diaphragm modelling mics inherently have a fixed amount of proximity effect. Dual diaphragm modelling mics at least have the potential to accurately model proximity effect. Differences in the accuracy of the dual diaphragm models represented here suggest differences in the accuracy of the modelling of the Sony C-800G’s proximity effect.
As important as it is, there is more to the sound of a microphone than just frequency response, for example, the difference in transient response between a dynamic and a condenser is arguably as significant as the difference in frequency response. Of course, all these mics are condensers and such differences should be minimal but it is worth pointing out that nothing other than frequency is measured here and there might well be other factors at play which aren’t reflected in the data presented.
Off-Axis Response
The off-axis response of microphones is seen as significant enough for Townsend Labs and Antelope to offer dual-diaphragm mics which can model the off-axis response of microphones. The Antelope data is a particularly interesting example as the Edge Solo is a single diaphragm mic and as a result, doesn’t model the off-axis contribution whereas the Edge Duo is dual diaphragm and the modelling is more accurate. This might be due to the use of a second diaphragm. The Slate system doesn’t use a second diaphragm.
Something else which is worth saying is that the Slate Virtual Microphone System originally launched as a system which included a dedicated preamp, the VMS One, the subsequently released VRS 8 interface features the same preamps. While the ML-1 can be used with any good quality preamp, the performance might well be more accurate using the Slate preamp. Assuming that Audio Test Kitchen used all of these mics into the same, non-proprietary preamps this might be a contributory factor into the variance in these results.
So there you have it, independent data presented without a conclusion. What do you think? Let us know in the comments section.