Re: Free Hands Friday #40 - A Song of Iron and Wire
I don't quite follow what one would expect to hear as far as the difference between a real and modeled amp for those incidental bumps on the guitar?
There seems to be quite a range of approaches to amp (or more generally analog) modeling in the industry, some very simple and some very complex. When we developed Tube Saturator at Wave Arts we realized there is a lot of marketing and not so much concrete detail about how different companies develop analog modeling - a tube amp could be modeled with something as basic as a lookup table that converts input to output samples based on a curve depicting the input to output voltage behavior of a simple triode amp. And many products probably use something like this to generate harmonic content similar to what happens when overdriving a tube, but something this simple won't be very accurate for larger systems with multiple triode stages and complex interactions of many analog components. So Bill Gardner (Wave Arts founder) came up with a sort of real time SPICE simulator (a program used to design analog circuits) that is solving a set of non-linear differential equations every sample period to accurately capture the grid current in a system of analog components, and this leads to very accurate audio output compared to the original circuit it's modeling. The downside is that it's exceptionally CPU intensive and is only practical for small circuits at this time (Tube Saturator is a simple Baxandall 3-band equalizer and a two stage triode preamp) - we didn't even attempt to deal with cabinets, power amps, and other things that contribute to the sound of a instrument rig.
We made a point of not shrouding this product in a bunch of marketing BS, Bill lays this all out in the TS manual:
http://wavearts.com/uploads/docs/TubeSa ... Manual.pdfThe point is that I'd be wary of dismissing digital amp modeling outright, when it's apparent there is a huge range of approaches to the problem and we may not be quite at the point where it's practical to model massive circuits down to the last intricate details, but we're getting closer. I'd be very curious to know what kind of approach Cliff at Fractal Audio is using for this amp modeling as it seems to be quite highly regarded. While I've used NI Guitar Rig a lot, I get the feeling that the DSP algorithms aren't exceptionally complex as they need to run many modules at once within a reasonable CPU limit.