Page 2 of 5

Re: The Ideas Collection

Posted: Sat Jan 26, 2008 4:42 pm
by Consul
Thank you for the links! SSE/SSE2 optimizations would be a good thing.

My interest in DSP actually started with wanting to make a high-quality guitar amp simulator, as the (in my opinion, anyway) best one on the market, by SofTube (http://softube.com/vintage_amp_room.asp) is kinda pricey and requires an iLok dongle. Now, I don't object to giving money for good software and to a good company (or person), but I do object to automatically being treated like a criminal. So the article on guitar and amp simulation in that second link is an interesting one to me.

This interest has now lead me to wanting to dive into circuit modeling of all kinds, so tools like CLAM and Faust will hopefully let me try out ideas quickly.

Re: The Ideas Collection

Posted: Sat Jan 26, 2008 6:14 pm
by cuse
Yeah I also thought about doing that. Because I also tried existing solutions and wasn't very impressed. I really would like to see my Boogie as virtual device one day as well, but in the meantime there's nothing better than the sound and smell of a real tube amp. :)

Re: The Ideas Collection

Posted: Sat Jan 26, 2008 6:33 pm
by Consul
That amp sim is based on a cubic waveshaper. I'm hoping for a more complete component model. I've been reading a lot of old electronics theory texts that describe how tubes work and how to model them. Granted, those models are for design purposes, but it's something to build on.

I've also been reading up on waveguide modeling (solving the wave equation in real time), and I'm wondering if I can apply that to the internals of tubes, transistors, and capacitors, as electrons are waves in a medium, after all. Maybe someone better at the math than me might be able to help me there. :) At the very least, I think waveguide acoustic modeling would be perfect for modeling the speakers and cabinets.

Re: The Ideas Collection

Posted: Sat Jan 26, 2008 9:28 pm
by sbenno
Hi Darren,
not sure if you already know this site:
http://www.musicdsp.org/

they have a collection DSP algorithms in form of code snippets and a mailing list (you should subscribe)
where you can discuss DSP related questions, I used it to read it alot in the past.
Many knowledgable people, many working for commercial companies too (both hardware DSP and software stuff
like VST plugins etc).

So I think this is currently the best mailing list for answers related to the simulation of a guitar amp etc.

Re: The Ideas Collection

Posted: Sat Jan 26, 2008 9:30 pm
by Consul
I'm actually already subscribed, thanks. I was just musing. I'll get back on topic.

Although it's not completely off-topic, as I'm hoping some good tube and other circuit component models will serve to make some nice-sounding filters and the like.

Re: The Ideas Collection

Posted: Sun Jan 27, 2008 6:52 am
by lowkey

Re: The Ideas Collection

Posted: Mon Jan 28, 2008 2:57 am
by Consul
Hehe, well, the Music-DSP list just told me that circuit modeling is treated as a "closely-guarded secret" by those who know how to do it, as it is their living. Nevertheless, I did get some links to papers and some tips. I'm tempted to reply back and mention my aim is to make an open source library. That'd probably guarantee no help coming my way. The thing is, everything points to this being differential equations, which I haven't taken yet, and won't until next Spring. I guess I keep reading and trying things out.

Re: The Ideas Collection

Posted: Mon Jan 28, 2008 12:23 pm
by cuse
Yeah, that's the difficulty with analogue circuits. You cannot simply create a model of the components of the electronic circuit and then calculate the total model by adding the results of each components like most simulation software software do. Or let's say in theory you could, but the required descretization time would be extremely small and the temporary values would have extreme values as well, so I doubt you can use this approach in practice.

Usually one creates a differential equation system for the model as a whole. But that's not easy. Literature about "feedback control systems" might be a good start for this field for you.

Re: The Ideas Collection

Posted: Mon Jan 28, 2008 3:32 pm
by sbenno
About those closely guarded secrets.
I remember when time stretching and pitch shifting was one of those magical forumulas only a few knew.
now open source apps like rubberband achieve more or less the same quality as commercial apps.

I think over time all those apps like guitar simularots will be available as open source providing comparable
quality as commercial ones.
But of course it takes a few brave men to make it happen so Darren I applaud your efforts and although the
journey will not be short it will be a rewarding one :)

as the old saying says ... information wants to be free :)

Re: The Ideas Collection

Posted: Mon Jan 28, 2008 5:16 pm
by Consul
The thing is, a completely discrete simulation must be at least possible within the computer, as there are real-time versions of Spice that can accept an input signal and process it. It's might not be efficient, but that would be the interesting problem to crack, right?

My dream is to have a library of virtual components that can accept samples as input over time plus static input parameters and then output what the component would do to the signal. Building a modeling plug-in of any kind then would be a matter of stringing together function calls in a way that mimics the circuit you want. There would likely be tweaking involved, but that's half the fun. :D

My calc teacher does constantly remind me that I do have Maple available at the school, and can use it to solve the complex math that I don't know how to do myself yet. ;)