This is why I love Pianoteq. It's a plug-in that virtually models a piano, so if you want 10 meter long strings, you can have that. Or you can go crazy with other parameters and get very strange pianos.
As someone who enjoys playing piano but doesn't have the space nor acoustic insulation for a real one, it's great being able to not just get the sound of different pianos whenever I want, but also virtually design my own. Modeling technology has really come a long way.
That's not to say that if I ever own a large enough house I wouldn't want an acoustic piano :-)
I second Pianoteq as a recommendation. It can run on Linux, so I have it on a Raspberry Pi attached to my digital piano. As a bonus, it also drives some addressable (programmable) LEDs: https://github.com/whyboris/Digital-Piano-LED
Thank you! Reading the original article, I was wondering if exactly this sort of physical modeling was possible or being done yet.
I'm still not sure what _sort_ of modeling it's doing (surely not full FEA of the entire vibrating system), but I'm sure we'll get there in another few years of GPGPU...
If I had to guess, they probably have a fairly accurate single-string model and then model energy coupling across strings. Another neat thing it does is it lets you position virtual microphones around the piano (and open and close the lid), so they definitely have full-instrument spatial awareness in their modeling.
One curious thing I noticed from the (simpler) string modeling synth in my keyboard is that if you up the dispersion it starts sounding like a bell... and indeed Pianoteq also has a very nice tubular bells patch. Explains what that's doing in a piano synth :)
As someone who enjoys playing piano but doesn't have the space nor acoustic insulation for a real one, it's great being able to not just get the sound of different pianos whenever I want, but also virtually design my own. Modeling technology has really come a long way.
That's not to say that if I ever own a large enough house I wouldn't want an acoustic piano :-)