Joker said:
So tell me what you think about this approach...
It does not work. I work with computer modeling and prediction of sound system performance, and I have seen the equivalent claims made for one of the software packages sold to that market. The claimers are either ignorant of - or willfully choose to ignore - some of the basics.
Claiming that a physical model implemented in a computer will produce a more accurate replica of the actual device's behavior than will taking
real data on the
actual device is downright silly on its face. Even
if the model were complete enough to account for what happens in the real world - it is not, and it will never be, in the case of acoustic wave propagation - the time required to enter the essential input data - which would have to include, as one component of many, every detail of every moving part in the speaker and all the materials used to make those parts - would be overwhelming for just one speaker/cab combo,
if you could actually acquire all that information (and you cannot).
In the modeling world, the benchmark for success is
how well the model predicts the behavior of the system being modeled. The way you quantify success is
to measure the response of the speaker you modeled and compare it to the response predicted by the model. But the
purpose of modeling in this case is to
avoid the use of the measured data.
The objections raised to the use of impulse responses are woefully misinformed. A properly-acquired IR of a speaker fully accounts for all the
linear effects they list, including internal reflections in the cabinet, port effects, nonideal transducer behaviors, diffraction from the cabinet edges, etc., in a way that
no implementable computer model will ever duplicate.
The way to get closer to the behavior of real cabs is to
improve the quality of the data you acquire on those cabs. An IR acquired by close-mic'ing will never produce the sound of the speaker as you hear it while you are playing through it. The required elements for acquisition of an authentic IR are:
1. Suitably neutral test mic. Here are some examples:
http://www.dpamicrophones.com/en/produc ... item=24012
http://www.acopacific.com/micdetal.html
http://www.earthworksaudio.com/25.html
These mics are quite expensive, but they add no audible coloration to the sound they receive.
2. Realistic choice of mic placement. Think of where your ears are relative to your favorite cab when you're playing. That's a good place to start. It won't be a few inches in front of the speaker cone.
3. Freedom from room reflections. We're interested in simulating a
cab here, not a room. You are
playing in a room. Adding some of the effects of another unfamiliar room will hurt, not help, in achieving the desired effect.
4. Adequate signal/noise ratio in the data taking process.
, but it sure is interesting to hear from a guy who's dealing with this sort of thing on a daily basis! Thanks Jay!
