RE: [buildcheapeeg] Software architecture

From: John Morrison (jmorrison_at_ahc.net.au)
Date: 2002-02-15 13:30:38


> 'Back-tracking' is hard for some types of DSP algorithms to handle,
> but easy for others. For instance, an IIR filter builds up a kind of
> memory of recent history in its buffers -- you can't just 'rewind the
> tape' and expect it to work straight off.
Hmm ok
So the rewind/analysis facility will only work with certain types.

> I think the best thing is for you to approach my analysis app as a
> kind of prototype. Once I have it working, and you've had a play with
> it, maybe you can see how it could be made to fit into your ideas. If
> it doesn't fit, maybe we could make 'analysis' and 'biofeedback' two
> different parts of the program ?
I'd prefer to make the one program/framework instead of wasted time in
parallel development.

> > > http://www.arts-project.org/
> > > http://kde.org/
> > Looks great BUT it's LINUX/KDE based and not portable at all. :-(
>
> The aRts layer is portable, but the graphical front-end is obviously
> KDE-specific. (Actually, as it is Qt based, perhaps it is portable to
> Qt/Windows -- I don't know)
OK I'll have to have another CLOSER look!

> > LADSPA sounds interesting and seems to be following the same lines as to
> > what I'd like our program to do.
> > I'll have a better read latter.......:-)
> > After all EEG signals are VERY much like audio signals. :-)
>
> LADSPA is a plugin standard (like VST). You need to look at some of
> the open-source apps that support LADSPA (like 'ecasound'). They are
> the ones that have the infrastructure and scheduling code already
> built and tested. If it's open-source and under a friendly licence,
> you can take what you need and adapt it as necessary.
Sounds like a GREAT idea to me. :-)
I hate to reinvent the wheel.

> Just one difference between audio and EEG, and that is that in
> processing audio, you often work with blocks of 64 or 256 samples at a
> time. With EEG, the sample rate is much lower, and you'd need much
> smaller blocks, or even no blocks at all (i.e. processing 1 sample at
> a time). Probably it is straightforward to configure existing code to
> work with smaller block sizes.
hmm I'll look into it.......that is why I wanted to look at it first before
Modifying existing technology.

> I hope you don't mind, but I plan to work assuming no modular system
> for now. However if you and others can do some research and find a
> good existing modular system that can be adapted and made to work
> well, then I'll certainly contribute DSP modules for it.
I don't mind you working assuming that the modular system doesn't exist. :-)
BUT why do you assume that we HAVE to use some other modular system?
I agree that we SHOULD and will borrow from other systems BUT it is my
belief that this is a unique situation and should be treated as such instead
of trying to shoe horn it into something else.
It sounds as if you know something of the modular side of things why don't
you share your knowledge?

> Also, if you have a look at what has already been done, you'll have an
> idea of the issues involved, and of the amount of work required if you
> decided to redesign and code it all from scratch.
I have no intentions of designing it all from scratch.
But doesn't it make sense to give it the attention this project deserves?

> At the moment, I'm just working with a "shortest path to where I want
> to go" kind of approach, and I don't personally need a modular plugin
> system yet, so I'll be putting my time into looking at DSP and things
> like that for now.
BUT if the modular system was available NOW, wouldn't it make your shortest
path so much shorter?
Great look into DSP and things! Personally I know nothing about them yet.

But I believe that the software side of things deserves just as much time
and thought as has gone into the hardware!
It is my belief that a system that allows users to work on the areas they
know best (DSP, User Interface, HW interface) and share that with all other
users. AND also allows them to pick the best of what is on offer, is the
best course of action for this.
The modular system won't be finished tomorrow or next week, in fact I hope
it's never finished I hope it's always improved upon. :-)

> Jim

John



This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:38 BST