Why do amp sims have to generate aliasing?

[earlevel]

It all kind of makes me wonder if digital is really the ultimate realm for audio...so, sure - one day fairly soon sample rates will probably routinely be so high that most artifacts are inaudible, but that still seems like doing it the hard way using brute strength.

And analog is...easy?

Let's see...

Mechanical...ever look inside a piano?

Electromechanical...ever look inside a Hammond organ?

Electronic...ever look inside an analog string machine?

Computer controlled analog...ever look inside an OBXa?

Fully computation digital synth...well, I have one on my iPhone; I can't say that looking inside is going to reveal anything simple or easy, but it sure is a lot easier to carry around...



But yeah, simply raising the system sample rate *is* brute strength, and that's why we don't do it that way (usually). Sure we do it that way internally for things like amp sims...for now anyway, until we figure out a more elegant way of doing the non-linear warping in a band-limited manner in one step.

But it's not that hard. I wrote Amp Farm 1.0 to run on the Digi hardware of back then—a slow 56002, not many cycles available—not a lot of brute strength. It wasn't easy, but it worked. Then DSPs got faster.

[earlevel]

...one day fairly soon sample rates will probably routinely be so high that most artifacts are inaudible, but that still seems like doing it the hard way using brute strength.

(Another comment, but first, I'm not commenting to bust your chops—I realize you're saying this semi-whimsically, and making a valid point...)

Think of oversampling as gaining frequency headroom. We can use a relatively narrow path (44.1 kHz for instance), but some time we need to do some intermediate work with a wider pipe before having the workflow re-enter the narrow utility pipe.

We do no differently with dynamic range headroom. Maybe our audio system is 16-bit, but for some processing the intermediate work needs to be in double precision (a filter, for instance).

[guitarzan]

...I wrote Amp Farm 1.0 to run on the Digi hardware of back then—a slow 56002, not many cycles available—not a lot of brute strength. It wasn't easy, but it worked. Then DSPs got faster.

Wow....you wrote Amp Farm 1.0? That was THE beginning of amp sims, wasn't it? That's great!!

Yeah, some digital amp sims do pretty good handling aliasing...I still like my first generation J-Station. It seems at least as smooth sounding as most current amp modeling software and it has better feel than most...I'm guessing multi-band compression somewhere...cab modeling maybe?

IMO cab simulation is where there is the most room for improvement. More dynamic for a better feel and fatter more complex tone. Any thoughts about that?

[earlevel]

Wow....you wrote Amp Farm 1.0? That was THE beginning of amp sims, wasn't it? That's great!!

Pretty much—certainly the first that anyone wanted to record and album with. Steinberg had Red Valve-It, which required a TDM DSP for the amp, and another for the cabinet as a separate plug-in (IIRC, only "combo" and "stack" choices on the cabinet, and they were pretty sad and didn't attempt to simulate any real cabinet—basically just a high freq rolloff). AF 1.0 had relatively simple (but better) IIR based cabinets (and much better amps). I think it was 3.0 and with newer Digi hardware ("HD"—56300-family DSPs) that let me go with better convolution-based cabinets based on newer Line 6 amps.

IMO cab simulation is where there is the most room for improvement. More dynamic for a better feel and fatter more complex tone. Any thoughts about that?

I tend to agree with that. I haven't had a chance to play with it myself, but if I were doing a lot recording and amp sim, I'd be tempted to pick up Audio Ease's Speak Phone 2.