The Mic Test Thread!

[teainthesahara]

Good points have been put forth that should help people figure out what they can discern, and what they can't from uncontrolled sound comparisons. However, such uncontrolled comparisons are not useless - the value lies 'within the noise'. Yup, they can be very limited if what one wants to hear is the result of a controlled comparison. But, there is rarely a post on this BBS that does not end off with:

YMMV (or any of its derivatives).

The strength of an uncontrolled sound comparisons is exactly that: YMMV takes on a true, tangible meaning - and that is worth hearing.

T

 

[Rhythmschism]

Hm, well it sounds like using impulse responses is a somewhat disputed subject, so I'll look into it further.

Sweetnubs, so, to answer my question, would you say that despite the scientific impossibility of reliably comparing or reviewing microphones, after gaining a sufficient amount of experience one is able to make an educated estimation of a microphone's character and quality? Would one then be able to communicate this estimation to others with any authority, or, would listening to the opinion of one that has sufficient experience still be "going out on a limb"?

 

[arcanemethods]

The idea of microphone impulses is a complete joke and another reason to sell people crap they don't need. You cannot record something with a radioshack electret condenser mic, use a elam 251 impulse, then make the source sound like it was recorded with a elam 251. You have lost the information that the 251 gathers by not using the 251.

Whatever you say. I won't try to disabuse you of this notion, at least not today.


Bob

 

[arcanemethods]

I dont think I have ever heard of the impulse response you refer to, it sounds very interesting. Is there any literature/internet reading you could point me to that would go into more depth on this? Perhaps this is the kind of standard that could create a standard that minimizes subjectivity. Thanks a bunch for your reply.

Sorry but there hasn't been much, if anything, published on the subject due the proprietary nature of success at it.

All I can say is that what is involved is pretty obvious to anyone well versed in DSP and Newtonian physics.


Bob

 

[Marik]

<The idea of microphone impulses is a complete joke and another reason to sell people crap they don't need.>

Sweetnubs,

Could you elaborate little bit more about this idea and why it is a complete joke? Also, considering that none of manufacturers ever published impulse response of mics I would like to know how it makes a reason to sell people crap.

 

[sweetnubs]

my good god man, I just gave a few reasons. I'll repeat them. We all know a sm57 peters out at about 17k if I remember correctly, could be a little lower. Now record a piano player playing in the upper octaves. You want to get all the high end frequency information of the piano correct? Now suppose you have an impulse for an earthworks mic which has a near ruler flat frequency response well into the higher frequencies which would capture this information. Render the impulse. It is not an accurate reproduction of the earthworks mic because all information above 17k was never recorded. The impulse cannot retreive information from thin air. Lets take this even a step further, take into consideration the sm57 is by design a dynamic mic which means its moving coil contains a lot of mass and is therefore less senstive to transients than the earthworks capictor based diaphragm. All of this transient information is lost in the INITIAL recording i.e. it was never there and therefore shall never be there. Run the Impulse. Is it truly emulating an earthworks mic since this frequency content and transient information is not there? Also crappy design introduces odd harmonic distortion which is very noticable and is one of the reasons cheap equipment sounds like crap. Record a source with a cheap mic which introduces IM distortion. This is now part of the program material. Run your Elam 251 impulse. The IM distortion is still there, does this represent a 251?

 

[arcanemethods]

my good god man, I just gave a few reasons. I'll repeat them. We all know a sm57 peters out at about 17k if I remember correctly, could be a little lower.

Depends on what you mean by peter out. If it goes to zero then you are right but it doesn't.

Impulse response methods can reclaim anything that is there at all, the caveat being that if it is too low then noise from either the mic itself or from the pre can become a problem. Your choice of the SM57 is about as bad as that gets except that being a dynamic means its self noise is _very_ low which means a really clean pre and you're golden even with an SM57.

Nearly any condenser will have more than enough response at all frequencies to morph that response to any other, both magnitude and phase, with no introduced artifacts.

Now record a piano player playing in the upper octaves. You want to get all the high end frequency information of the piano correct? Now suppose you have an impulse for an earthworks mic which has a near ruler flat frequency response well into the higher frequencies which would capture this information. Render the impulse. It is not an accurate reproduction of the earthworks mic because all information above 17k was never recorded.

Bad assumption. It's down but its there.

The impulse cannot retreive information from thin air. Lets take this even a step further, take into consideration the sm57 is by design a dynamic mic which means its moving coil contains a lot of mass and is therefore less senstive to transients than the earthworks capictor based diaphragm. All of this transient information is lost in the INITIAL recording i.e. it was never there and therefore shall never be there.

Again, not gone. Just down. All of the transient information is in the impulse response just at perhaps lower levels.

Run the Impulse. Is it truly emulating an earthworks mic since this frequency content and transient information is not there?

What makes you so sure its not there?

Also crappy design introduces odd harmonic distortion which is very noticable and is one of the reasons cheap equipment sounds like crap. Record a source with a cheap mic which introduces IM distortion. This is now part of the program material. Run your Elam 251 impulse. The IM distortion is still there, does this represent a 251?

Now you've hit on the one thing impulse response methods cannot do. They cannot capture any distortion that may exist in the emulated mic and cannot change that of the actual recording mic. OTOH, any mic today that has signifigant, or even audible, non-linear distortion would be laughed off the market. Mic distortion is a mere fraction of the distortion of the very best speakers and even the best headphones. Not a problem.


Bob

 

[sweetnubs]

To sum up an emulation cannot add information that never existed or remove information "encoded" onto the source material by the original microphone. Nor can a mic emulation emulate the proper application of the emulated mic relative to the orginal source material. For example I would choose a certian preamp for my c414eb for the particular source in which I like to apply the c414eb. I would position this mic relative to the source in the way I know the source will sound its "best" considering the particular preamp being used, the interaction of the source the room I am very familiar with, etc. and how the mic and preamp will interact as a single unit with the source and the room and source as a single element. This placement also takes into factor the recording medium. I may select a different mic, mic preamp, positioning based on the fact I may record to analog tape at +9. All of this requires "before the emulation" knowledge. Now if you recorded this source with a '57 and try to emulate the 414eb it will ultimately fail because part of the equation needed to successfully emulate the 414eb is this "before the emulation" knowledge. Let's say for arguments sake your 414eb emulation is 100% accurate, what you've in essence done is successfully emulated a badly placed 414eb paired with the wrong preamp recorded to the wrong medium. ps. Anyone who is aware of the characterstic sound of intermodulation distoration is well aware of the fact that IM distortion is easily audible in cheap gear. I hear it all the time. Christ I heard a rode NT1 that sounded like nails screeching on a blackboard. Once you are aware of this sound cheap gear will constantly make your skin crawl.

 

[DJL]

Is that why mic emulators suck eggs when compared to the real mics?

 

[Marik]

<Marik, what is a microphone impulse response chart and how is it useful? Thanks>

DJL,

I think Bob can answer this question much better than I possibly could, so... Bob, could you elaborate, please?

Sweetnubs,

I think the confusion comes from the fact that you are reading into Bob's post too much. You put all together--frequency response of SM57, impulse of Earthworks, IM distortions of cheap mics, and emulation of Ela M251.
For some reason for many years mic manufacturers give only static specs of the mics. However, dynamic characteristics are much more important. Although I cannot speak for Bob, but all he has said in his post is that impulse response (which is a dynamic function) much better represents what is actually going on in the mic, at least that's how I read it. Is a given mic actually suits particular situation? That's completely different thing?

Feel free to correct me if I am wrong.

<Anyone who is aware of the characterstic sound of intermodulation distoration is well aware of the fact that IM distortion is easily audible in cheap gear. I hear it all the time. Christ I heard a rode NT1 that sounded like nails screeching on a blackboard. Once you are aware of this sound cheap gear will constantly make your skin crawl.>

Are you sure you hear IM distortions here? And how do you know?

 

[arcanemethods]

To sum up an emulation cannot add information that never existed or remove information "encoded" onto the source material by the original microphone.

To sum up, you are being insuficiently analytic. One can of course treat the whole gestalt as a single indivisable unit and say you can't get there from here and I'll agree with you. You can, however, take more of an engineering approach and consider the decomposition of that gestalt into its component elements and treat them separately. That may not work well in this new age but it works pretty well in practice.

You have indirectly pointed out another limitation of emulation. It will only give really predictable results between microphones having the same pattern. If you record with a cardiod and try to emulate a figure 8, it of course won't change the pickup pattern but it will still quite substantially give you the space the cardiod sees with the basic personality and color of the figure 8.

This mix and match can even be thought of as a feature of emulation. No reason a mic has to physically exist to sound pretty damned good for a particular application.

You may have had bad experience with distorting mics but I'd call such mics broken. I've yet to find a reasonable mic that I could measure the distortion of because all I could get out of it was that of the speaker which swamped it.

I've actually done a lot of work with this and developed and applied the DSP techniques involved so my opinion on the matter is at least based on a good deal of research and listening to the results.

That research is aimed at a product. One of the big holdups is that I only trust measurements I make myself in a product that will have my name on it . I'd sell my soul to have been on the first team that was noted in the start of this thread so that I could disappear for ten minutes with each of those mics.


Bob

 

[sweetnubs]

some of the reasons. Don't get me going on the fact the emulation is digital. You mean I gotta convert to digital to emulate a mic? Can't remove that. What track to digital? Are you freakin' nuts. We ain't there yet.

 

[sweetnubs]

c'mon pal, don't be coy, you know you can hear odd-harmonic distortion brought upon by fast, cheap and bad designs. It's subtle but there. Record a mandolin sometime. I think I do not need to be overanaylitic. You can't take away, you can't add what's not there and you cannot emulate positioning and signal chain for the would-be emulated mic and now you've taken an analog source and made a digital representation of it. There's still more, shall I continue . . .

 

[arcanemethods]

<Marik, what is a microphone impulse response chart and how is it useful? Thanks>

DJL,

I think Bob can answer this question much better than I possibly could, so... Bob, could you elaborate, please?

The impulse response is the output that the mic would give if the stimulus it was given was an extremely brief pulse, in digital terms it would be one sample wide and no more for a given sample rate. The signal that the mic puts out in response to that impulse is, oddly enough, called its impulse response.

Contained in that is all of the information relative to the frequency response of the microphone when stimulated by a source at the position of the impulse generator. The most common would be on axis but with a directional mic, two on axis measurements at different distances will disclose its proximity effect and one should take measurements at a few other angles to do other kinds of magic with them.

These things aren't any good to look at, only a really trained eye will be able to tell much at all and that's not much but if you do a Fourier transform of that impulse response, you get the magnitude and phase of its frequency response across the spectrum, part of which you are used to seeing approximations to on data sheets.

In the most general and unqualified terms, if you know the impulse respones of two mics you can, with DSP techniques, make either sound like the other. The two processed together in a certain way will yield another impulse response that can be used via convolution to transform one to the other.

You don't actually measure these things with impulses, but rather with sinusoidal sweeps (or known noise patterns) and then apply a DSP process to the recording of that sweep that converts it into the impulse response.

Elaborate enough?

Although I cannot speak for Bob, but all he has said in his post is that impulse response (which is a dynamic function) much better represents what is actually going on in the mic, at least that's how I read it. Is a given mic actually suits particular situation? That's completely different thing?

That's certainly true. All the emulation does is transform the frequency domain coloration of one mic to another. It doesn't change patterns nor does it do anything about distortion. I still maintain that distortion of modern mics, even cheap ones, is not worth considering unless it is broken.


Bob

 

[arcanemethods]

c'mon pal, don't be coy, you know you can hear odd-harmonic distortion brought upon by fast, cheap and bad designs. It's subtle but there. Record a mandolin sometime. I think I do not need to be overanaylitic. You can't take away, you can't add what's not there and you cannot emulate positioning and signal chain for the would-be emulated mic and now you've taken an analog source and made a digital representation of it. There's still more, shall I continue . . .

Nope. It wouldn't be fruitful.


Bob

 

[sweetnubs]

allow me anyway. Suppose I was recording a trumpet with a live band all playing together in the control room ('cause they want to) and I use a rode nt1 (god I hate that mic) on the trumpet. Now in this instance I would normally select perhaps an RE20 because it would reject the other instruments much better than the NT1 condenser. Now let's say I want to emulate the Re20. It won't work because now the phasing problems associated with using the NT1are now imprinted on the source and cannot be removed. This will not "represent" an re20 used in this situation even though you modified it's frequency response with an impulse file. So all have to do is buy an sm-57, a Fairchild impulse and a Frank Sinatra Impulse and a Carnegie hall Impulse and now I can sound like Frank Sinantra being compressed by a fairchild in Carnegie hall all with an sm57? hmmmmmm . . .something sound fishy kids.

 

[Marik]

<The impulse response is the output that the mic would give if the stimulus it was given was an extremely brief pulse, in digital terms it would be one sample wide and no more for a given sample rate. The signal that the mic puts out in response to that impulse is, oddly enough, called its impulse response.

Contained in that is all of the information relative to the frequency response of the microphone when stimulated by a source at the position of the impulse generator. The most common would be on axis but with a directional mic, two on axis measurements at different distances will disclose its proximity effect and one should take measurements at a few other angles to do other kinds of magic with them.

These things aren't any good to look at, only a really trained eye will be able to tell much at all and that's not much but if you do a Fourier transform of that impulse response, you get the magnitude and phase of its frequency response across the spectrum, part of which you are used to seeing approximations to on data sheets.

In the most general and unqualified terms, if you know the impulse respones of two mics you can, with DSP techniques, make either sound like the other. The two processed together in a certain way will yield another impulse response that can be used via convolution to transform one to the other.

You don't actually measure these things with impulses, but rather with sinusoidal sweeps (or known noise patterns) and then apply a DSP process to the recording of that sweep that converts it into the impulse response.

Elaborate enough? >

Beautifull!!! Thanks Bob.

<you know you can hear odd-harmonic distortion brought upon by fast, cheap and bad designs.>

Sweetnubs,

I have been in music since 9 months before I was born and in audio for only about 25 years. I have designed and built all kind of equipment--speakers, amplifiers (mostly tube), tape recorders, line, phono and mic pres, turntables, DACs, mics of almost all kinds and configurations...
Some say that I have pretty good ears. But you know, to be honest, I never can say where are harmonic distortions, where are IM ones, where are odd, and where are even ('more pleasant') ones. But ones again, I am still new to it...
To me, to talk about you hear odd-harmonic distortions in such a low signal level device as a mic, sounds little bit odd, especially considering the fact that most of these mics are actually clones of very good designs. Most likely you hear bad quality parts. Right now I am working on MXL603 modification and can tell that mere change of parts for better quality ones makes significant difference. Has the amount of odd-harmonics changed?--no way. Has sound changed?--I will post clips later and you tell me.

Oh yeah, one more thing. I always talk about things I have experienced myself, or know about for sure. If not, I prefer to shut up and listen what more knowledgable people have to say.
Am I sceptical about what Bob talks about mic simulation? May be.... But I will refrain from any judgement until I try and see it myself.... 'cauze it seems that Bob knows what he is talking 'bout.

 

[Marik]

Oh yeah, BTW,

Put into any stock cheap Chinese mic some degree of negative feedback and you will get absolutely negligable amount of any kind of distortions, including odd-harmonics, and it will sound... like complete crap

 

[DJL]

The impulse response is the output that the mic would give if the stimulus it was given was an extremely brief pulse, in digital terms it would be one sample wide and no more for a given sample rate. The signal that the mic puts out in response to that impulse is, oddly enough, called its impulse response.

Contained in that is all of the information relative to the frequency response of the microphone when stimulated by a source at the position of the impulse generator. The most common would be on axis but with a directional mic, two on axis measurements at different distances will disclose its proximity effect and one should take measurements at a few other angles to do other kinds of magic with them.

These things aren't any good to look at, only a really trained eye will be able to tell much at all and that's not much but if you do a Fourier transform of that impulse response, you get the magnitude and phase of its frequency response across the spectrum, part of which you are used to seeing approximations to on data sheets.

In the most general and unqualified terms, if you know the impulse respones of two mics you can, with DSP techniques, make either sound like the other. The two processed together in a certain way will yield another impulse response that can be used via convolution to transform one to the other.

You don't actually measure these things with impulses, but rather with sinusoidal sweeps (or known noise patterns) and then apply a DSP process to the recording of that sweep that converts it into the impulse response.

Elaborate enough?


That's certainly true. All the emulation does is transform the frequency domain coloration of one mic to another. It doesn't change patterns nor does it do anything about distortion. I still maintain that distortion of modern mics, even cheap ones, is not worth considering unless it is broken.


Bob

Thanks Bob... but I've yet to hear a mic or amp modeler/emulator that didn't suck eggs when compared to the real thing (but it can be a fun effect)... why is that? Thanks

EDIT ADDED: PS... and can a trained eye really tell that much from an impulse response chart or would it really take a computer to analyze it? Thanks again.

 

[arcanemethods]

Oh yeah, BTW,

Put into any stock cheap Chinese mic some degree of negative feedback and you will get absolutely negligable amount of any kind of distortions, including odd-harmonics, and it will sound... like complete crap

I don't follow this, Marik. You can put the front end FET in a feedback loop by using a source follower circuit, which most will, but that only eliminates the FET distortions. I know of no way to put the physical capsule in a feedback loop and I'm puzzled by your statement that cleaning it up would make it sound like crap.

BTW, harmonic and intermodulation distortion are two sides of the same coin. Any non-linearity that causes one will also cause the other. The nature of the non-linearity will affect the spectral distribution of the distortion products but you can't have one form of distortion without the other. There are types of non-linearities that make the harmonic distortion more or less benign but the IM distortion that results will still be offensive regardless of the nature of the harmonic distortion. Their magnitudes are pretty much proportional to commom factors but the harmonic can be clustered at frequencies relative to the fundamental that are less offensive. No such property exists for IM.

As far as diaphragm distortion there is no difference between a cheap and an expensive mic. That distortion is a result of the fundamental physics of stretched diaphragms and electrostatics. There is not really any room to wiggle in this regard in a design. The tension, polarization and diaphragm to plate distance completely determine it.

One big exception to this is the differrence between electrically and electret polarized designs and the electret is the winner in this regard. If the charge difference between diaphragm and backplate can migrate radially as the diaphragm moves in and out, it will and a non-linearity in the response results. With an electret the charge is embedded in a dialectric and cannot move.



Bob