The Fremer lays an ostrich egg thread
- Thread starter DaveyF
- Start date
No that's not it at all. Acousticians do measure acoustics using an impulse of sound. A common source was a blank fired from a starter's pistol. One problem with that is that the explosion isn't necessarily the same from gun to gun or even from one shot to another from the same gun. Also it probably doesn't eminate as a spherical waveform, the shock wave probably travels cylindrically or conically as the expanding hot gas follows the trajectory of the shell. Today many use a dodecahedral array of speakers. Whatever method they use to generate an impulse, it will be impossible to segregate its components by frequency. To do that, a quasi white noise generator called an ILG fan is used. It's known spectrum is compared with that measured at some distance at various points. The results, that is the differences are presented as a band within which lies most of the steady state spectral transfer functions for the room. Not good enough and it also doesn't address the vector nature of the resulting field. Usually an omnidirectional microphone is used for measurement although one scheme uses 6 omnis, one at each end of a boom on three intersecting axies.
The measurement scheme that follows the model is nothing like that at all. It measures vectors by virtue of a spherical array of unidirectional microphones. The test signal which can have any arbitrary intensity as a function of angle of propagation (useful for emulating different musical instruments that radiate differently) due to using a spherical array of loudspeakers exploits a simple trick in calculus that allows echoes to be measured one frequency at a time. By assembling data for multiple frequencies across the audible passband, each reflection's relative spectrum as a function of time, relative intensity, and angle of incidence on the microphone array compared to the first arriving sound wave can be mapped. IMO it was the only really clever part of the whole thing, the rest seemed straighforward. There aren't really any tricks to it, it was just seeing that you could look at sound from a different point of view by considering just the fields at only two points at a time. It seems nobody had ever done it that way before.
Modern way to measure impulse response is with a swept sine and inverse FFT. It solves the problem of needing a very loud transient as you mention. This is how REW works by the way.
s
Thanks Amir. I'll have to look into it so I can see how it works.
Have you got any references indicating how this is applied to measuring room acoustics?
Bruce B
WBF Founding Member, Pro Audio Production Member
This is also what I was talking about using maximum length sequences like white/pink noise and de-convolution via FFT instead of the transient impulse responses.
My pleasure. Here are some papers in my library. A couple are available online:
"Simultaneous measurement of impulse response and distortion with a swept-sine technique", http://www.google.com/url?sa=t&rct=...1IGIDQ&usg=AFQjCNGNgpkubVSaMY_Opydf1mhq6XZOvQ
"Transfer Function Measurement with Sweeps": http://www.google.com/url?sa=t&rct=...wIH4Ag&usg=AFQjCNFZ9Xuqcny8gtRGKF-WDDaZrFK4jw
And this AES papers: "Effects of Oversampling on SNR Using Swept-Sine Analysis"
While all this scientific talk is fascinating to the layman, such as myself, I would like to ask a layman's question to SM.
What is the difference between the system that SM is discussing, and the commonly used 'amp modeler/ cabinet modeler' that has been in wide usage in the pro-audio field for several years now?
The amp modeler essentially uses various digital re-creations of the sound of various amps/cabinets/ rooms/halls/sound effects and allows the user to tailor the sound to one's desire at the output to the PA or amplifier/speakers.
Is this what SM is striving for in the home environment? If so, I for one can see the potential in the technology.
What is the difference between the system that SM is discussing, and the commonly used 'amp modeler/ cabinet modeler' that has been in wide usage in the pro-audio field for several years now?
The amp modeler essentially uses various digital re-creations of the sound of various amps/cabinets/ rooms/halls/sound effects and allows the user to tailor the sound to one's desire at the output to the PA or amplifier/speakers.
Is this what SM is striving for in the home environment? If so, I for one can see the potential in the technology.
Thank you for the references Amir. It will take me some time for me to read and understand them. Perhaps after we could discuss them. I'm looking to see whether this system is capable of characterising the spectral content or relative frequency transfer functions of individual reflections. This was one of the requisites of my model and test instrument.
As I understand them, these modelers generally can get one amplifier to sound like another. Insofar as their ability to model and recreate room acoustics, compared to my model they are crude, inaccurate, and entirely inadequate. To give you an idea of why, if you pick up a book about acoustics and just thumb through it, you will see starting with the inventor of the science, Wallace Sabine and to this very day, acoustic scientists have come up with one parameter after another, a large number and variety of different parameters to describe mostly how rooms affect sound. When you try to put them together you get a confused jumble of notions, a tapestry that is sometimes incoherent and self contradictory, taken together incomprehensible. Beranek's papers comparing concert halls and opera houses trying to find a correlation between measurements and what golden ears like is a kind of admission that the science is still at the groping stage of development.
An analogy is like the story of the three blind men sitting in front of an elephant. Each can touch one part of the elephant and describe it to the other two. One has hold of the tail, one the trunk, and the third a leg. Among the tree of them, none have any real idea of what an elephant is about or looks like. Now imagine that you stood there with your eyes open able to see the entire elephant from all angles in color and 3 dimensions. You now have some idea of how to make a statue of an elephant that looks something like the real thing. That's what my model does. It gets to look at sound from an entirely different point of view that makes it possible to understand it. It unjumbles the spaghetti of knots. Each of their parameters is now explainable and understandable from this better and complete single perspective. The sound systems based on it and the measuring device merely reflect this different understanding.
Last edited:
They started out as simple amp modelers, but have evolved very rapidly. I own one of the first generation Pods (the first commercial guitar amp modeler) and a very current modeling amp from Fender (a Mustang III). We're talking guitar amplifiers, of course. These days, they don't just model amps, they model the response of those amps, the effects in those amps, the speakers in (and not in) those amps, and more. And the best of them are very, very good at it.
The magic of tube guitar amps is the bane of hifi -- clipping. In a guitar amp, it can turn the amplifier into an extension of the musical instrument, giving the player dynamic control over "overdrive" or soft clipping, just through the strength of his attack on the strings. The tone of tubes is also pretty fabulous for guitars. I've never played anything but tubes outside of the house until this year, when I bought the Mustang. My beloved vintage Fender Deluxe Reverb is in retirement.
With this digital modeling amp, I can not only choose from a very fine collection of amp models, allowing me to play a clean Fender Twin Reverb for a rhythm part and switch to an overdriven Vox AC-30 Top Boost for the solo, but I can create combinations that never existed -- a Deluxe (a 1X12 combo) playing through a 4X12 Mashall cabinet, for example. This not only changes the the tone, it changes the load the Deluxe is playing into, altering its response to effects, to tone and volume adjustments on the guitar and the amp and even to my playing technique. I can take it a step further and play a Vox through that Marshall 4X12 cabinet, with the reverb circuit from a vintage Fender reverb unit. I can go a step even further and adjust the virtual bias of the amp to give it more or less "sag" -- another thing that is badd in hifi but can be useful in a guitar amp. I can place effects before the amp or in the FX loop between pre and power, and the impact it has on tone is exactly the same as it is with physical placement of actual effects pedals. I suppose it's hard to understand how powerful all this is if you're not a player. I plug my amp into my laptop, load up the software, manipulate models, save them as presets, and then play as if I have a whole collection of vintage amps on stage with instantly switchable adjustments. It's more like magic than technology.
This is not simple stuff. These modeling amps control an incredible amount of nuance, and I have no doubt that this technology could be very successfully applied to hifi and even to environmental acoustic modeling. But it would take significant R&D, and would be targeted to a market that thinks tone controls are a bad thing and digital processing is the spawn of the devil.
I won't be holding my breath.
Tim
Thanks Tim, that is a great description of what an amp modeler can do. I agree with you, I do not think these devices are simplistic in any way. I use a Vox Tonelab, which even goes as far as to use a tube in the circuit to 'assist' with the sound. The parameters that are accessible and adjustable are numerous with these devices, my Tonelab can model numerous different amps/cabs/halls/pedals/etc, and IMO it sounds fairly close to the original. . Plus, I do think that they may hold the next significant step forward in the home audio environment.
SM, I am still very confused as to what it is that you are using that does all of this 'processing' and allows you to manipluate the sound so much, that you can say it is like 'looking at the whole elephant'. Are you basically using a device that is akin to the Lyngdorf? Digital manipulation of the speakers to emulate a different tone and therefore modify how the speaker sounds in different room environments?
And you do well. We are still in the point where measurements can not fully describe the sound of our best amplifiers or other electronic devices. Why should this market you are referring to trust in a new product that has not even be designed?
Anyway, I hope you accept that there is a large difference between a guitar amplifier and an high-end amplifier.
No, it doesn't do any of that. The method does the same thing the gun shut method do. Except that it doesn't need anything that loud yet produces even better signal to noise ratio. With a single microphone, you are always measuring the whole room as a black box.
The method to measure single reflections that i have seen use an anechoic chamber and then use another speaker to play that "reflection." Studies have been done this way with large number of reflections.
I was a little confused by Davey's question because i have a general understanding of amp modelers (as well as a point to point wired Marshall and a repro Vox) and I never understood the modelers to deal with room acoustics, more the combo of amp head, speaker cabinet and various pedal/distortion/reverb effects. Tim, your description simply confirms that. Am I missing something here? These modelers don't really pretend to emulate room acoustics, more the sound of the amps themselves, right?
Davey, not being pissy, just trying to get a grip.
Bill, I'm not sure about Tim's particular modeler, but I believe that some of the latest Line 6 models can emulate various rooms and halls..Probably as the particular amp being modelled would sound in those environments. So, yes, that's not really dealing with room acoustics in that sense.
Nonetheless, I would think that this ability could be added using DQX as an example.
Which brings me to my next question to SM....couldn't you get the same effect by simply using an equalizer, or DQX
Bruce B
WBF Founding Member, Pro Audio Production Member
And this goes along with all the hardware emulation plugins. There are plugins that model all the famous Teletronix, Pultec and UA as well as newer hardware like the Manley Massive Passive and such.
And you do well. We are still in the point where measurements can not fully describe the sound of our best amplifiers or other electronic devices. Why should this market you are referring to trust in a new product that has not even be designed?
Anyway, I hope you accept that there is a large difference between a guitar amplifier and an high-end amplifier.
Oh, I don't know, there are guitar amps out there built into gorgeous furniture, point-to-point hand wired (covered with beautiful cloth in spite of being inside the amp), silver soldered, filled with vintage NOS tubes and custom transformers, etc, etc...it's all very much like "high-end." But I will concede that it's very different from hifi. And I think you've missed the point; if there were a market for the products to justify the R&D, the engineers who produce these devices and the ones Bruce talks about that model classic studio gear could not only measure and fully describe the sound of your amplifiers, they could create convincing copies of them. They could put a dozen of them in a single box and allow you to virtually adjust the input impedance, output impedance, tube swap, you name it. I'll bet they could even make it sound like you changed boutique cables. Point and click synergy, baby. Now, it would be a lot more expensive for you guys than it is for a simple guitarist such as myself, because I just need midrange and blistering volume. Hifi needs full range and, for many of you, huge scale. So you'll still need big iron for big speakers. Big enough to deliver the full impact of any model in the box into any load you can reasonably anticipate. But none of that is a limitation of the modeling itself. That could be done. If they can make a digital model and a class D amp respond dynamically, to my touch -- to my pick attack and finger vibrato -- just like a 50 year old class A, hand-wired, all tube amp responds, I think they can navigate the gap between a ML and a Dart Zeel without breaking a sweat. The high-end just needs better engineers.
Tim