|
Termyn8or -> RE: Acoustic Signature Recognition and IDENTIFICATION (9/28/2007 8:24:48 AM)
|
thorn, I was actually talking about mechanical resonances.
These are transducers, and if they use a diaprhram they resonate at some frequency(s). Every speaker and microphone in the world does.
Electrical resonances can be dealt with by proper loading and damping networks, but in a microphone, very little can be done about a mechanical resonance. In a speaker though, mechanical resonance is abated somewhat by an amplifier with a high damping factor. This is alot harder to apply to a transducer that goes the opposite way, that is movement to electronics, rather than electronics to movement.
Quite a while ago, gawd am I old, I was forced to connect a ceramic phono cartridge to a magnetic cartridge input. Now a ceramic inherently is designed to cancel the RIAA equalization curve and provide a (somewhat) flat response to a line level input.
I had envisioned a network, possibly even needing to be active to accomplish this, but further investigation revaled that since the element of the transducer was cheifly capacitive, I might have a quick and easy way to do it. With a potentiometer I loaded the output and at a certain point not only was the level reduced to match the sensitivity of the phono preamp, the frequency response also flattened. I think the ultimate load was about 2K ohms. The sound was quite good, crisp and clear, and I would say most likely better perfomance overall than if the cartridge had been plugged into a normal, line level input.
But see this is all electrical. With this FFT technique, what you are measuring is mechanical resonance(s), and that can be masked by inherent resonances in the transducer (the mike).
Perfection is impossible, but they can come close, and close costs money. Sometimes it happens almost by accident, or someone really does have a better idea. I bought my speakers for a bit over MSRP when they were ten years old, and I wouldn't sell them for ten times that amount. They seem to have no resonance(s) at all.
What alot of people do not realize is that our brain does a form of FFT all the time. Ever hear your own voice on a tape recorder, I mean a really good one ? Still doesn't sound right does it ? The reason is as simple as it is complex.
To draw a quick analogy, if you walk out of a purple room with purple lights, everything outside will look green. This effect abates as your brain adjusts to the new environment.
When you walk into a room there is always sound, no matter how quiet. The room itself resonates. Your brain automatically applies equalization to what you hear and it sounds natural. There is also the fact that your vocal sounds are also transmitted to your ear through your bones, but that is a constant and is not addressed here. It is not the point.
To illustrate this for yourself, take a piece of cardboard, and a speaker grill. Hold each about two inches from your face and speak. Notice the difference ? But that is an expected difference. Your brain compensates nearly immediately.
Also, each person's ears have a different frequency response, even with what is considered normal hearing, you have been living with those specific "transducers" all your life. That explains why different people set the tone controls or EQ differently on their audio equipment.
In an FFT analysis, a true mechanical resonance cannot be removed in the analog domain. The FFT program must be calibrated to it. This is done mathematically in the digital domain. Even if the transducer's response has been flattened in the analog domain, there will need to be a gamma correction for a certain frequency range.
For the electronically inclined, if you like to screw around with it, get your hands on a square wave generator, feed it's output to a ten band equalizer and view the output on an oscilloscope. As you adjust the EQ controls, take note of how it affects the square wave. This can give you a rough idea of how FFT works. A very rough idea, but something tangible. It is almost as if you can see a frequency response curve, in reverse, on the peaks of the not so square wave.
This is what an FFT actually does, but it is much bigger, better, bouncier and all that.
The single tick, FFT method of measuring a speaker's frequency response was also touted to reject the acoustic proprties of the room where the test was performed. Previous to this method it had to be done in what is called an anechoic chamber. The theory being that if any room is big enough, the measurement is done before the reflected sound reaches the mike. Sound theory actually, as long as the room is at least 15' by 15' or so. The measurement is actually completed in about 0.025 seconds or less. Any input after that time is disregarded.
I'd like to digress here into speakers for a moment. When they started making alot of cheap large speakers, most of them had ports. Mainly because they didn't have the balls to seal the cabinet, because it is so flimsy. Porting also reduces cabinet resonances. However, that's when they really started to cheat. Not only were they using pink noise, and an untuned transducer, which counts harmonic distortion as valid output. See when you have cheap speakers and put 25Hz into them, you get a little 25Hz, but you also get alot of 50, 75, 100, all the multiples. This is called harmonic distortion in the view of any audiophile.
Well they really did it when they started using the close mike method. Now they add the woofer cone's oiutput to the port output. That is so wrong. First of all they are out of phase, they cancel out to some extent. What's more a ported speaker's response varies with the environment in which it is used to a much greater extent than an infinite baffle (sealed) design.
This is all germaine to the subject of using acoustics and FFT to ID things. Even the relative humidity of the air must be compensated for, but this is one of the easier parts. It is not much harder than dealing with a "red shift" or a "blue shift" in spectrography.
I still don't see a hell of alot of use for this except in military or security applications. Of course those seem to be growing fields. Hmmm, I am tired of my job(s), maybe I should look into this.
T
|
|
|
|
