"Where does the tone come from in a microphone?"

[GeorgeToledo]

This is simply not true, you are breaking laws of physics here. Can you provide any documentation to substantiate this? Or measurement? If two microphones have same exact frequency response, and their upper and lower limit of FR extends to the same point they will have exactly same TR. It seems you are mixing up TR and IR.

However if you have two mics that have same exact FR say 20-20.000hz but FR starts to differ above 20.000hz you do indeed get different TR.

I don't get why make things so complicated. Transient response is determined by the frequency response extension, or bandwidth. You can not limit FR and have good TR, same as you can't have wide FR and claim TR is bad just because the mic sounds dark.

I am of course talking about operation in linear region, and presuming microphones are not clipping.

Assuming that which is defined to not work linearly, does work linearly, is probably not a good foundation.

Varying harmonic distortion with level isn’t necessarily recognized as clipping, but is enough to make the situation more complex.

This is the same reason the amp modeling market has developed more complex techniques than using a single IR.

A mic would have to have impulse response taken at various levels, on axis, off axis, etc, and then have some interpolation scheme that would at best be an approximation, in order to have some sort of *rough* mathematical model.

Textbook examples of minimum phase devices vs non, immediately reveal several common sub-systems in microphones to not be minimum phase by definition.

 

[Co.]

This is simply not true, you are breaking laws of physics here. Can you provide any documentation to substantiate this? Or measurement? If two microphones have same exact frequency response, and their upper and lower limit of FR extends to the same point they will have exactly same TR. It seems you are mixing up TR and IR.

However if you have two mics that have same exact FR say 20-20.000hz but FR starts to differ above 20.000hz you do indeed get different TR.

I don't get why make things so complicated. Transient response is determined by the frequency response extension, or bandwidth. You can not limit FR and have good TR, same as you can't have wide FR and claim TR is bad just because the mic sounds dark.

I am of course talking about operation in linear region, and presuming microphones are not clipping.

So let's compare a ribbon mic with a severe HF roll off but super snappy and clear transient with slowish, smearish LDC with huge peaks around 5k and 12k.
Are you basically saying, that I am going to get the exact same transient response, if I just EQ them to have the exact same frequency response?
So with the right EQ, I am going to make the LDC less smeary and faster? And darker at the same time? (I'd love to have an EQ like that!)
Probably by cutting a lot between 2-20k and doing crazy boosts above 20k? How can I boost something, that is not there?
I not trying to prove a point. My technical knowledge is certainly not on the same level as yours. I just want to get the concept.

 

[metalockpick]

A transducer has size, mass, excursion, and Q at various frequencies. Ideally (for the best specs) you want the mass to be zero, the size to be infinitesimally small, the excursion to be infinite and the Q to be very low for accurate frequency reproduction. Obviously, that's not the case, so there are trade-offs in reality. That adds to the complexity.
If the frequency testing is not done with impulses/FFT, You don't really know what you're seeing. It could be noise energy unrelated to the actual input signal.,Also if you compare the frequency response of a LDC with a 6 micron diaphragm and one with a 2.5 diaphragm, perhaps one that spent 20 years having the gold flaked off of it being ahead of a kick drum, you will see that the reduction in mass, while increasing the high end, does not necessarily increase the high end extension due to the physical size of the diaphragm.
King, you have seen microphones with similar frequency response on axis, that are very different off axis. This can be due to mechanical properties with the transducer or acoustic properties with the housing. So, yes. In a general sense, depending on the testing mode, you can have two devices with the same "frequency response" that have different impulse response/transient response, and they will sound wildly different.

 

[kingkorg]

So let's compare a ribbon mic with a severe HF roll off but super snappy and clear transient with slowish, smearish LDC with huge peaks around 5k and 12k.
Are you basically saying, that I am going to get the exact same transient response, if I just EQ them to have the exact same frequency response?
So with the right EQ, I am going to make the LDC less smeary and faster? And darker at the same time? (I'd love to have an EQ like that!)
Probably by cutting a lot between 2-20k and doing crazy boosts above 20k? How can I boost something, that is not there?
I not trying to prove a point. My technical knowledge is certainly not on the same level as yours. I just want to get the concept.

You can't look at things that way. First of all the ribbon would, i presume, be figure of 8. LDC would be cardioid. Apples and oranges, you'll never get the same response no matter what you do, neither FR or TR.

But leave that aside, it's an extreme example and it wouldn't make any sense anyways to try to match condenser and ribbon microphone unless they have same pattern.

You are right, you can't recover something that isn't there, but you can remove what is there with an eq or filter.

If you had in theory same pattern ribbon and condenser. Ribbon had a roll off starting at say 15K, and condenser extending to 30K. You could make a filter that matches ribbon HF roll off and apply it to the condenser and they would have the same TR. Same goes for low end. What would happen at 10K would have nothing to do with transient response, that is just amplitude part of the signal at that frequency.

Transient response has to do with extension or bandwidth of the frequency response, and nothing to do with shape of it.

Here you have c800g which will sound brighter and "quicker" but it's response falls above 15K, and knowing k67 it will continue to droop above 20K. It will have inferior TR compared to Earthworks that will sound darker and "slower", but it's FR extends to 30K.

 

[kingkorg]

Here's an example of image that creates confusion and makes me crazy. The way it is often explained is that part that I marked on the ribbon represents the slow transient response and the ribbon flapping around after the initial transient.

This is simply wrong. The part I marked is the part of transient response that indicates ribbon's sensitivity to very low frequencies. Additionally the ribbon is F8 and it has substantial boost at LF due to the proximity effect which is larger than cardioid. We can deduct that the ribbon actually has better TR at very low frequencies.

The part I marked with yellow does show superior TR, but you have to ask what is the upper frequency of the signal, it could very well be 10K where the mic has a boost, while the dynamic and ribbon could have a dip, so the image doesn't proove anything without providing details about how the test is performed. It is useless.

 

[soliloqueen]

I think a lot of this is due to a conflict between theory and reality. It is technically possible to restore transient response or correct phase delay using EQ but you would need to accommodate for every single element in the sound in reverse and every element of the sound would have to be affected by being captured the same way. Extremely tight laboratory conditions would be required to produce a recording where you could do this and prove it.

There's also just no such thing as "The frequency response" of a microphone or even "the polar pattern." It's all a big undifferentiated blob of picking up sound and these measurements are cross-sections of a hugely multidimensional response. We take measurements of the response in specific places so that we might be able to compare them to other measurements taken in specific places and infer but at the end of the day these things are more useful for quality control than anything else, which is why that's precisely what they're used for by most reasonable engineers.

 

[kingkorg]

I think a lot of this is due to a conflict between theory and reality. It is technically possible to restore transient response or correct phase delay using EQ but you would need to accommodate for every single element in the sound in reverse and every element of the sound would have to be affected by being captured the same way. Extremely tight laboratory conditions would be required to produce a recording where you could do this and prove it.

Exactly. Again the confusion comes from marketing and advertising, presenters babbling about stuff they know nothing about

 

[kingkorg]

There's also just no such thing as "The frequency response" of a microphone or even "the polar pattern." It's all a big undifferentiated blob of picking up sound and these measurements are cross-sections of a hugely multidimensional response. We take measurements of the response in specific places so that we might be able to compare them to other measurements taken in specific places and infer but at the end of the day these things are more useful for quality control than anything else, which is why that's precisely what they're used for by most reasonable engineers.

Also true. Being fortunate to measure so many microphones I came to conclusion most people like smooth variations in the frequency response without crazy sharp dips and boosts. Moderate HF boost if any. Translates well to how people use "musical sounding" EQs - think Pultec. Sharp things at specific frequency are hit and miss, and can work with some sources, think SM57... These often point to unwanted resonances within mic.

Off axis response is where all the woodoo, magic, vibe, 3d is. Probably impossible to quantify, control, correct, manipulate by any scientific method or measurement. It ends up the way it is and it either works or doesn't.

 

[metalockpick]

Here's an example of image that creates confusion and makes me crazy. The way it is often explained is that part that I marked on the ribbon represents the slow transient response and the ribbon flapping around after the initial transient.

This is simply wrong. The part I marked is the part of transient response that indicates ribbon's sensitivity to very low frequencies. Additionally the ribbon is F8 and it has substantial boost at LF due to the proximity effect which is larger than cardioid. We can deduct that the ribbon actually has better TR at very low frequencies.

The part I marked with yellow does show superior TR, but you have to ask what is the upper frequency of the signal, it could very well be 10K where the mic has a boost, while the dynamic and ribbon could have a dip, so the image doesn't proove anything without providing details about how the test is performed. It is useless.

This is excellent. Part of the reason why the ribbon doesn't have the frequency response extension could be that the physical size of the ribbon means that you get phase cancellation @ high frequencies.
I commend you on providing graphs that weren't created by the marketing department (!). With all the phase cancellation though, I'm not sure that you could simply use EQ to get the high end back, even though the ribbon may be able to move fast enough.

 

[Co.]

You can't look at things that way. First of all the ribbon would, i presume, be figure of 8. LDC would be cardioid. Apples and oranges, you'll never get the same response no matter what you do, neither FR or TR.

But leave that aside, it's an extreme example and it wouldn't make any sense anyways to try to match condenser and ribbon microphone unless they have same pattern.

You are right, you can't recover something that isn't there, but you can remove what is there with an eq or filter.

If you had in theory same pattern ribbon and condenser. Ribbon had a roll off starting at say 15K, and condenser extending to 30K. You could make a filter that matches ribbon HF roll off and apply it to the condenser and they would have the same TR. Same goes for low end. What would happen at 10K would have nothing to do with transient response, that is just amplitude part of the signal at that frequency.

Transient response has to do with extension or bandwidth of the frequency response, and nothing to do with shape of it.

Here you have c800g which will sound brighter and "quicker" but it's response falls above 15K, and knowing k67 it will continue to droop above 20K. It will have inferior TR compared to Earthworks that will sound darker and "slower", but it's FR extends to 30K.

Well, I actually find the transient representation (to use a less technical term) often more accurate in a ribbon, compared to most LDCs. Possibly because ribbons actually extend higher than LDCs, if you take the amplifier section into the equation. At least that is what my experience is, when recording with high sample rates and slowing the recording down 2-4 octaves. With ribbons this usually sounds pretty impressive and realistic (as realistic as something can sound, being slowed down that much). With your typical LDC, this sounds really bad, really fast.
So I'm not doing measurements. I'm more interested in how I can work with that stuff and come to my (probably faulty) conclusions by listening.

 

[kingkorg]

This is excellent. Part of the reason why the ribbon doesn't have the frequency response extension could be that the physical size of the ribbon means that you get phase cancellation @ high frequencies.
I commend you on providing graphs that weren't created by the marketing department (!). With all the phase cancellation though, I'm not sure that you could simply use EQ to get the high end back, even though the ribbon may be able to move fast enough.

Even if you could bring back what is lacking you would bring back some noise that wasn't there in the first place. And then we come back to noise issue. Like everything else in life - it's complicated.


@Co. This is very specific situation, and i'd say just use whatever works best, it comes down to that.

"At least that is what my experience is, when recording with high sample rates and slowing the recording down 2-4 octaves."

You bring up very important point regarding the sample rate and slowing down, it's often overlooked when it comes to high sample rate advantages.

 

[Clarkster]

The Electronics

 

[Barry Hufker]

I hope thls post somehow adds to this discussion and is not a distraction from it.

Andy Simpson (Simpson Microphones) once told me (and I'm paraphrasing from statements made years ago so please don't hold him or me too accountable...) You never see a microphone company post the phase response of their microphones, and for good reason. It tends to be horrible.

To change that, Andy designed his own microphone which aimed for best phase response. This gave it an unusual frequency response that you would compensate for in post-production by using (phase linear?) EQ. The microphone pattern was very directional, which made it difficult to use in stereo for instance, but I've never heard such clarity/transient response from anything else that comes close to his microphones. Comparing a recording of an orchestra I made with his mics and some high quality conventional mics (B&K, Schoeps), it was a real revelation as to what is possible. I could hear very clearly, very deeply to the back of the orchestra in great detail in a way the other mics couldn't convey. And the whack from percussion was stunning in its realism.

This is a discussion of the mics from way back when. Andy took a lot of heat for his ideas. Andy Simpson's New Microphone
Attached are two pictures of his microphone, profile and then inside the horn.

 

[Icantthinkofaname]

with slowish, smearish LDC with huge peaks around 5k and 12k.

What does "smearish" even mean? Every time I see somebody describe a condenser mic as smeary it seems to mean "this mic is too dark for my tastes".

 

[Co.]

What does "smearish" even mean? Every time I see somebody describe a condenser mic as smeary it seems to mean "this mic is too dark for my tastes".

I wouldn't describe it as dark. I'd say it means "unclear", coming from a resonance.
Can be good or bad, depending on taste and application.
A neve EQ is probably more "unclear" than a Calrec, but this can be nice.
Weird membrane resonances, that are filtered out with not so great electronics in a condenser mic are "unclear" in a way that probably nobody likes.

 

[abbey road d enfer]

Textbook examples of minimum phase devices vs non, immediately reveal several common sub-systems in microphones to not be minimum phase by definition.

Mathematically, linearity is a prerequisite for MP.
It seems to confirm that MP in audio is a slightly different concept that the academic one.
Most litt seems to confirm that MP is recognized when it's good enough for a definite domain.
Since it involves limitations due to personal choices or measurement set-up, the conclusions, particularly regarding the bijective relationship between frequency and phase responses are to be taken with those restrictions in mind.

 

[abbey road d enfer]

So let's compare a ribbon mic with a severe HF roll off but super snappy and clear transient with slowish, smearish LDC with huge peaks around 5k and 12k.
Are you basically saying, that I am going to get the exact same transient response, if I just EQ them to have the exact same frequency response?
So with the right EQ, I am going to make the LDC less smeary and faster? And darker at the same time?

"Faster" and "darker" do not go hand-in-hand. But do you really hear "fast" or "slow"? Audition does not perceive "speed" per se, it just analyses the fequenc content.

(I'd love to have an EQ like that!)

You may find that such an EQ has limited practical utility, because EQ is static and frequency response varies dynamically.

Probably by cutting a lot between 2-20k and doing crazy boosts above 20k? How can I boost something, that is not there?

Something that's not there does not exist in this case. response at 20kHz may be 50dB down, it's always possible to boost it electronically to make it linear. Of course, IF the system is Minimum-Phase.
And there'll be noise.

 

[gyraf]

@Barry Hufker - it seems that the Simpson was what Marik refers to as a "Backloaded" ribbon motor - the horn is about matching acoustic impedance to try to counteract the ribbon's natural HF rolloff. EDIT: Nope - it was a low-sensitivity dynamic (read on in this thread for details)

Simpson website went out around 2016: https://web.archive.org/web/20120712235539/http://www.simpsonmicrophones.com/

 

[abbey road d enfer]

I wouldn't describe it as dark. I'd say it means "unclear", coming from a resonance.
Can be good or bad, depending on taste and application.
A neve EQ is probably more "unclear" than a Calrec, but this can be nice.
Weird membrane resonances, that are filtered out with not so great electronics in a condenser mic are "unclear" in a way that probably nobody likes.

Smear has a precise sense, which is the displacement of harmonics. Any peak or dip in the frequency response results in phase-shift, which is exctly that. It moves the harmonics by a certain amount, early if the slope of the frequency response is ascending, and late if the slope is descending. The sharper the peak or dip, the higher the slope, and the higher the phase-shift.
Let's take cymbals, that have a large and relatively flat spectrum, and a mic that peaks at 10kHz, the frequencies below 10k will be moved earler and those after 10k moved later.
Now there may be other causes for phase-shift, like diaphragm break-up, which produces phase inversion and mutiple harmonics, or multiple acoustic paths.
The audibility of "smear" is not a simple phenomenon, because the audition does not know phase directly, it knows it by the effects on frequency response, and it detects smear as annoying when it produces certain effects on frequency response.
As ricardo wrote, Minimum-Phase is how God wanted it, so the audition is satisfied when the effects of smear are coherent with frequency response.
Smear is bad when the displacement of harmonics does not concur with the frequency response effects.

 

[MagnetoSound]

Hi Barry, I remember the discussion on the Simpson very well. I remember too finding the principles quite fascinating. Sadly, the world was not ready for such a convoluted and cumbersome design (that relied on corrective eq but did not provide it), and at such an outrageous price.

@gyraf - the transducer was a dynamic diaphragm, I believe, not ribbon. I may be wrong. Also, having read several discussions again, I think the acoustic impedance principle was about eliminating, or reducing, IM distortion, not about FR.