The Virtual Microphone

[SSLtech]

So long story short you can gat a similar sound for limited conditions, but there is no way ever to change a mic's pick up  pattern after the fact "in the mix".

While I agree largely with the spirit of the post, this little section is not technically accurate, insofar as the SoundField microphones can be polar-pattern and axial-angle adjusted after the fact.

In fact I'd go even further than you do John; even ON-AXIS a mic can pick up differently, because of pressure wave versus velocity wave.

A theoretical demonstration: Take two extremely different microphones: for the sake of this example let's consider a dipole mic such as a simple ribbon, and take a pressure mic such as a sealed capsule mic. (envisage a tin can with the top removed, and cling-film stretched over the opening. Now imagine how varying external pressure makes the film dip in and bulge out.)

Put the two mics in the exact middle of a length of plastic pipe maybe eight inches diameter. Now seal each end with an 8" loudspeaker drive unit. Wire the two speakers in series, "in-phase". -Now when a tone is played through the speakers, BOTH speakers will move inwards together, and outwards together. The pipe will pressurize and depressurize. The Omni capsule (the bean-can) will hear lots of signal. the velocity at the middle of the tube however, is zero. That means that the velocity mic  (the 'ribbon') will 'hear' nothing.

Now flip the polarity of one speaker, so that when one pushes air into the pipe, the one at the other end pulls air out of the pipe. -Now we have a slug of air moving along the pipe and back again, but never increasing or decreasing in pressure. The Ribbon mic produces a fine signal, but the omni (pressure) mic produces no signal...

Now, how does that play out in real life? -Well, different instruments produce sounds differently, but let's start by considering a simple, infinite baffle (completely sealed) speaker cabinet. The speaker cone swings in and out, alternately pressurizing and depressurizing the air inside the cabinet, and conversely depressurizing and pressurizing the air outside the cabinet. It's a simple pressure variance, though effected through what is effectively a piston surface.

Some speaker designs however use porting, or veltilation like the simple bass-reflex design. -The bass reflex addresses the issue of the low frequency rolloff by venting a 'timed' wave from the REAR of the cone to 'assist' the faltering response driven by the front. -The result -if done right- is held to be an extension of an octave of flat response before the low frequency rolloff, but then the response falls off TWICE as steeply. -You do get TWO octaves of positive boost 'for free' before the tradeoff goes negative.

The combination of a pressure wave from the front of the cone face and a velocity wave from the port means that an omni mic and a dipole mic will record VERY different frequency/amplitude responses, even ON-AXIS, in an anechoic chamber, even using perfect;y "flat" microphones (if you can find a satisfactorily flat ribbon!)

Okay, so what else APART from a speaker cabinet? -Pretty much everything in fact. -The human voice is an excellent example where most of the 'guided/channeled' air coming from the mouth is velocity wave, but sibilants/fricatives are pressure. (Anyone thinking that Chinese mics sounding "spitty" in a manner where simple EQ doesn't solve the problem sounds familiar right about now?)

"Mmmmm" sounds and chest resonance is largely pressure wave, "aaaaaaaaa" is largely velocity wave... and the balance between component waves varies across the spectrum.

...So good luck to anyone trying to pick THAT little puzzle apart. -Yes it's possible. -But you'd have to start with a mic like a Soudfield, or -at the very least- a Josephson C700a.

Then perhaps all on and off-axis patterns could be modeled, along with pressure/velocity correction. -But not before.

That -in a long-winded way of explaining- is why the 'Royer simulation' in the video is laughably impossible.

 

[JohnRoberts]

It's nice to see a thoughtful and more comprehensive discussion of this, my short post was incomplete.

Since there are actual mic designers hanging out here I won't embarrass myself by lecturing to them, but of course there are other differences too.

Given enough conditional information (data about the mic being modeled, and data about the sound source** being captured), and enough computer power I expect you could get close. As I already mentioned I kicked this tire and moved on back in the mid '80s as too hard to do well (especially back then).

Today when mics remain one of the few things on our audio chains that actually should sound different, maybe this is predicting a generic mic with presets and knobs so you can cover old standards or, go where no mic has ever gone before (wherever that might be). 

I have nothing further of substance to contribute to this discussion.

JR

@ Steven Slate... welcome to the party. It's always good to hear from the horse's mouth what a product's design goals are. 
 
** I know from years of working closely with drums that they do not make a simple sound field. As I recall from reading an old (very old) audio book, different acoustical musical instruments vary more than microphones for their sound transmission patterns. 

 

[Steven Slate]

Thank you John for the kind welcome.

I'll echo that from our tests, it was very clear that the M7 and CK12 are very unique sounding capsules, and of course it should come to no surprise that every M7 in itself, was unique sounding after a 50 year aging.

My point was about the cardioid polar pattern and the relationship of the deviation of frequency response in the off axis.  In both the M7 and CK12, we found that the off axis response had similar ratio of frequency alteration to the on axis.

Furthermore, even when the algorithm was about 70-80% convincing on a close mic scenario on axis, we found that the off axis was much harder to discern from our reference mic.  This was good news as we feared during the development that it would be extremely difficult to get a convincing algorithm both on and off axis.

With the smaller SDC microphone, we have a much bigger challenge.  To get from the very flat and quick condenser element tone to a dynamic tone (such as that of the SM57) is actually a much larger challenge then getting the large diaphragm to sound like another large diaphragm mic.  One of the first things that occurs is a lot of transient manipulation, followed by a severely dynamic filter algorithm that triggers a very specific harmonic reaction.  So yes, modeling a $79 microphone turned out to be much worse than the $30,000 vintage classic.

I've been browsing this forum for quite some time, thanks again for having me, I've learned a lot from the folks here.

Cheers,
Steven

 

[riggler]

Does your DSP remove the non-linearities / character / color from your 2 source mics? (ML1 and ML2) --- to get to a known neutral?

(Then I am assuming the DSP processes the selected model...)

Or are your models programmed against the signature of the source mics?
And on that note, I would gather you must hold some pretty high tolerances in the manufacture of the mics, for consistency.

The more I see the Raven in use the more I like it. I watched the video on the mic system and you said: "Skeptical? You should be!".

This is the right place! I would like to try these with my orchestra rig out of curiosity!

 

[ruairioflaherty]

The most obvious difference between the CK12 capsule and Neumann’s M7 used in the early U47s is that the former used an edge-terminated dual backplate design, as opposed to Neumann’s centre-terminated capsule. However, hidden inside the backplate, the AKG design operates in a fundamentally different way from the Neumann one, although the differences are technical and subtle: the former is based on a resonator design and the latter on the ‘aperiodic’ concept. The dimensions of the diaphragm to backplate spacing and the chamber labyrinths within the backplates also evolved over the life of the CK12 design, in an effort to increase its sensitivity. Amusingly, the side-effect of this tinkering was to boost the extreme high-end response, which subsequently became an integral characteristic of the C12’s infamous sound. Also, early C12s employed 10-micron-thick Styroflex diaphragms, but later versions (and all the Telefunken and Siemens models) used lighter and more responsive 9- and latterly 6-micron Mylar diaphragms.

Pucho,

You may want to credit where you got that.... - http://www.soundonsound.com/sos/may10/articles/peluso22251.htm

 

[pucho812]

your absolutely right. lack of sleep on my part. Plus they did do a good job of esplain it so why reinvent the wheel and retype an explanation.

 

[pucho812]

I wonder what Gus would say about this.

 

[Steven Slate]

Does your DSP remove the non-linearities / character / color from your 2 source mics? (ML1 and ML2) --- to get to a known neutral?

(Then I am assuming the DSP processes the selected model...)

Or are your models programmed against the signature of the source mics?
And on that note, I would gather you must hold some pretty high tolerances in the manufacture of the mics, for consistency.

The more I see the Raven in use the more I like it. I watched the video on the mic system and you said: "Skeptical? You should be!".

This is the right place! I would like to try these with my orchestra rig out of curiosity!

Hi Riggler. You are correct in that the first algorithm is meant to zero out the entire signal.  However, due to good hardware design, this task is quite simple.  Whatever remaining artifacts are left from the capture are very linear and predictable.. unlike the mics that we have been modeling which are the complete opposite.

Indeed, we have to have very strict tolerances on the hardware and are in the process of developing a QC process that is efficient and effective.

Cheers,
Steven

 

[pucho812]

What voltage are you polarizing you mic capsule at? Is the mic distortion free?

 

[emrr]

Nice discussion, good to see SS chiming in. 

 

[pucho812]

guess i drove away slate

 

[Steven Slate]

Hi pucho. I'm happy to report that it will take a bit more than copying an SOS article or asking me about the voltage of polarization of the capsule to drive me away.

On an earlier prototype I recall an 80v polarization but that may have changed. I'll discuss with the engineering team and will gladly report back.

Cheers,
Steven

 

[pucho812]

awesome. Thanks.  As far as copying sos, sure, why not, they were so eloquent in their words that I could not have said that information any better however I could of added a few minor non important details such as which capsules went to which mic based on testing at the factory but that would be irreverent to the discussion. 

 

[pucho812]

btw 80V seems fairly high

 

[SSLtech]

btw 80V seems fairly high

Not at all... for an SDC, it certainly isn't!

I have four U47's here right now; two originals and two modern recreations (hell, a few weeks ago, I had five). They make a nice sound, but they can't discriminate between axial versus non-axial pickup, nor the pressure-versus-velocity ratio even before we consider the spectral range for these considerations. Given these truths, there's a limit to how much 'simulation' should be claimed. The pressure-to-velocity component will likely change between the center and the edge of a loudspeaker cone in many cabinets, as well as with distance, possibly to differing degrees between infinte-baffle cabinets (such as 'Marshall cabs' versus open-backed combos like Fender Twin Reverbs) so there's a relevance to consider.

While it's a tantalizing goal to 'recreate' classic (and expensive) rarities, the limits of possibility have to be considered.

That's not to say that this product doesn't sound good... but you know that home studio owners are going to simplify this down to "it's exactly like a vintage mic".

 

[Steven Slate]

Again, if we're talking about using the mics for the purpose of recording sound, my answer is this:

Duplicating the off axis response has proven to be far easier than duplicating the on axis close response.  My proof of this is that when we had the on axis response to about 80%, we could not tell the difference in the off axis response.  We have a few theories about why this is the case, but anyway, it is.

I'd be glad to post a few links to some files when I'm back in the office next week.

Either way, this is a forum of technologically wise professionals who I'm sure know full well the depths in which we've achieved in digital, and therefore can appreciate it. A friend who works in the robotics field is about to publish a paper on his company's artificially intelligent nanobot processor that is the size of a dime.  To him, making a digital model of frequency, harmonics, saturation, and phase is laughably simplistic.

Cheers,
Steven

 

[pucho812]

then maybe he should be designing your stuff ;D

 

[ricardo]

My point was about the cardioid polar pattern and the relationship of the deviation of frequency response in the off axis.  In both the M7 and CK12, we found that the off axis response had similar ratio of frequency alteration to the on axis.

Furthermore, even when the algorithm was about 70-80% convincing on a close mic scenario on axis, we found that the off axis was much harder to discern from our reference mic.  This was good news as we feared during the development that it would be extremely difficult to get a convincing algorithm both on and off axis.

Steven, having had some success at trying to make one speaker sound like another, I fully appreciate your achievement.

It's certainly easier when the capsules are about the same size.

But I have a question about "algorithm was about 70-80% convincing on a close mic scenario on axis".

I take it this is a Listening Test.  Do you do this with the 2 mikes in an Anechoic, a Sound Booth, a nice sounding hall .. or all three?  Source?  Vocals?
___________________

I'd dearly like to attempt replicating the sound of U47, U87, C12 etc with TetraMic & VVMic Technology.  Just need someone to lend me Golden Pinnae Engineer certified examples of same. 

What SSLtech describes is essentially how I confirmed the Velocity & Pressure response of the Soundfield in Jurassic times.

I'm also not sure about C12s ever having other than polyester (Mylar) diaphragms.  This is from Bernhard Weingartner who designed the beast for AKG and was a personal friend of Clem Beaumont, my mike mentor at Calrec.

 

[Steven Slate]

We started out using anechoic chambers with various test files and sources and very quickly found out that it just doesn't work.  We needed to combine those tests with many more real world environments and sources.  After about a year we finally got a test procedure, some which I obviously can't reveal, which seems to really paint a pretty solid picture of what we need to do with each mic's unique algorithm.

Cheers,
Steven

 

[SSLtech]

I have to be careful about replying to this without coming across as dismissive, however this forum is a more focused on the technical aspects than most, and the last reply was very vague... -I 'get' not revealing proprietary information, but -to quote "Scotty" (James Doohan) from Star Trek:

"Ye cannae change the laws of physics"

Not even with sales-talk.

Again, I don't doubt the cleverness of the product, or the people involved... but the laws of physics suggest that the information collected by these microphones is insufficient to execute a more complete simulation... no matter HOW immensely clever the people involved.

..unless someone has repealed the laws, and I somehow missed the memo.