KVR Audio

detfella · Post by **detfella** » Mon Feb 09, 2004 8:32 pm

Do people generally use matlab for researching and development of plug-ins before they are implemented in C++ code?

blümchen · Post by **blümchen** » Mon Feb 09, 2004 9:10 pm

Yes, all of them.

And they also study 6 years scientific algebra before they are authorized to release audio plugins.

Markleford · Post by **Markleford** » Mon Feb 09, 2004 9:17 pm

agent 007 wrote:And they also study 6 years scientific algebra before they are authorized to release audio plugins.

Tell me about it! I'm currently studying for the VST Certification Exam. Lots of late nights ahead...

- m

detfella · Post by **detfella** » Fri Feb 13, 2004 2:49 pm

detfella · Post by **detfella** » Tue Feb 17, 2004 2:16 am

not too sure where u were both going with your replies....It was a serious question.

I also wondered if it was possible to use converted matlab code (into C++) for use within a vst. Or is this not allowed due to mathworks licensing agreements?

blümchen · Post by **blümchen** » Wed Feb 18, 2004 3:49 am

detfella wrote:not too sure where u were both going with your replies....It was a serious question.

I also wondered if it was possible to use converted matlab code (into C++) for use within a vst. Or is this not allowed due to mathworks licensing agreements?

Of course I was kidding.

I personally think, that most of the devs here don't use MatLab for some reasons.

But who can forbid you to use CONVERTED MatLab files?
They can merely forbid You to use MatLab output directly...

And if you have MatLab, so you can look at the licensing agreement for such questionable limitations...
(This would be the reason for not using MatLab at all.)

For what to f*** is MatLab finally good for, if you cannot use the results in any way than for pure imaginary research or highly academic orgasms?

Have they copyrighted the usage of maths in general

I have seen at least one open source project, where the initiator used MatLab output files directly inside the code. But I cannot remember where...
I think, it was a kind of BLIT oscillator...

Take a look at Music DSP Source Code Archive (see the links section here on kvr).

Urs · Post by **Urs** » Wed Feb 18, 2004 11:21 am

Hmm,

the only scientific tool I use is gnuplot...

I usually do a lot of mutations of new algorithms in C and decide the way to go by listening to them.

Urs

texture · Post by **texture** » Wed Feb 18, 2004 12:44 pm

Urs wrote:the only scientific tool I use is gnuplot...

I've been looking for something like this - cheers!

blümchen · Post by **blümchen** » Wed Feb 18, 2004 1:28 pm

Right.
Finally plugin audio technology is made for musicans.
More intuition please!

I have seen alot of "scientific" or "academic" source codes and plugins (see CSound or STK or wathever). Errrm...

In contradiction to this, the best plugins these days come from inspirated musicans and hobbyists having no clue from "high society math".

whyterabbyt · Post by **whyterabbyt** » Wed Feb 18, 2004 1:38 pm

agent 007 quoth In contradiction to this, the best plugins these days come from inspirated musicans and hobbyists having no clue from "high society math"

So where do you think the core algorithms for implementing filters came from? Random chance?

Urs · Post by **Urs** » Wed Feb 18, 2004 1:41 pm

whyterabbyt wrote:So where do you think the core algorithms for implementing filters came from? Random chance?

hahaha...

true, the bilinear z-Transform is something one should know. But one doesn't need MatLab to make its outcome sound good :b

Urs

whyterabbyt · Post by **whyterabbyt** » Wed Feb 18, 2004 1:46 pm

Well that's still the difference between a good programmer with a good ear 'tuning' (for want of a better term) an existing 'high society math' algorithm, and actually inventing new algorithms.

IIR, FIR, FM, physical modelling (including waveguides), digital reverb, phase vocoders, et.c et.c - all 'high society math' at the end of the day.

Urs · Post by **Urs** » Wed Feb 18, 2004 2:07 pm

whyterabbyt wrote:IIR, FIR, FM, physical modelling (including waveguides), digital reverb, phase vocoders, et.c et.c - all 'high society math' at the end of the day.

Sorry, but some of these are a bit too trivial to be called 'high society'...

For instance, FIRs and Waveguides are sooooo obvious concepts - I intuitively built them long before I knew their names. On the other hand, knowing what to do with them can be less trivial.

Digital Reverb is a good example where Math rarely helps. Most reverb algorithms seem to come from long trial and error sessions, in which the correct ratio of delay times had to be found... Only ears apply in that respect...

Well, and though understanding FFT is highly non-trivial, using it is quite simple...

But... DSF or WaveLet... yeah... hard stuff. Not used very often...

Urs

whyterabbyt · Post by **whyterabbyt** » Wed Feb 18, 2004 2:18 pm

Sorry, but some of these are a bit too trivial to be called 'high society'...

His term, not mine. But all of them have been researched and documented by academics

For instance, FIRs and Waveguides are sooooo obvious concepts - I intuitively built them long before I knew their names. On the other hand, knowing what to do with them can be less trivial.

They might be obvious once one has a knowledge of other DSP algorithms, but would one really wake up one morning and think 'I know I'll wack some numbers sequentially through an array of coefficients and see what happens' if one didnt

Digital Reverb is a good example where Math rarely helps.

Yeah, but it took a couple of academics (Moorer, Schroeder) to start the ball rolling.

Most reverb algorithms seem to come from long trial and error sessions, in which the correct ratio of delay times had to be found... Only ears apply in that respect...

Surely thats tuning of the implementation of a concept, not creating the concept itself.

Well, and though understanding FFT is highly non-trivial, using it is quite simple...

Yeah; get that plugin from DelayDots and go...

But... DSF or WaveLet... yeah... hard stuff. Not used very often...

tobybear · Post by **tobybear** » Fri Feb 20, 2004 8:38 am

Just a side note:
Anyone not interested in buying MatLab might want to know that there is a free alternative called "Octave" (http://www.octave.org). It is compatible to MatLab on the basic level, some toolbox ports also exists, it runs fine on Linux, Windows and whatever and, yes, you can use it for DSP prototyping

Cheers

Toby

www.tobybear.de

R&D