Re: [Audacity-devel] Re LV2 (was GSoC Update)

[Lars L. <lar...@gm...>]

Sorry, this got lost somewhere in my inbox...


On Thu, 2008-03-13 at 17:33 +0000, James Crook wrote:
> Lars Luthman wrote:
> 
> > There's the LV2 Vocoder (https://gna.org/projects/lv2vocoder)
> 
> But that's a repackaging of Josh Green's LADSPA wrapper around Achim 
> Settelmeier's Vocoder?  So already available under LADSPA?

Ah, so it is. Missed that.

> > the Calf plugins...  ...will most likely be released before the 
> > summer, according to author...
> 
> Remember, we're looking to find something that showcases some benefit of 
> LV2 over LADSPA.  Is there (will there be) a Calf plug-in that fits that 
> bill?

Well, the synth plugins definitely. 

> > and some of my own plugins...
> 
> Picking one of those may be the best route to your showcasing the new 
> features.  If there's a particular one you have in mind, do mention what 
> it does on your application and it can be one of the targets.

I'll have a look.

> Our Tone effect and Chirp effect use the same source code for generating 
> the audio data.  The tone effect shows a subset of the available 
> parameters.  Chirp sets the pitch at two points, Tone at one.  A 
> generalised Chirp could have parameters for each 'set point'.  From the 
> GUI creation point of view adding set points would be analogous to 
> adding voices...

Ah, yes - I think I misunderstood what you meant by the Tone/Chirp
effect in relation to the dynparams extension. Yes, adding envelope
points or something similar and having new parameters pop up for every
point sounds very much like what the dynparams extension is for.

> It piqued my interest because more complex effects have many parameters. 
>   That's not only intimidating to new users, it also can mean that you 
> keep switching parameters back and forwards when using the effect in its 
> full 'advanced' mode - when you're using the one effect in two different 
> ways.

In LV2 this wouldn't require the dynparams extension though, it would be
enough with something much simpler - you could just use some RDF
predicate to set the "priority" or "importance" of different controls.
Controls with priority 0 would always be shown, controls with higher
priority could be shown in "advanced mode" (if needed there could be
arbitrarily many levels of complexity). I know that Dave Robillard has
been talking about having something like that for his Ingen modular
synth as well.

> I've edited most of the text away, but I generally like the sub goals 
> you set out.
> 
> This one's a bit sticky though:
> > * Once the simple tone generation using LV2 synths is done, it would be
> > easy to extend it to generate chords or even short note phrases, for
> > example using a tiny piano roll editor in the plugin dialog. It wouldn't
> > be used for serious composition, more for things like NHL-organ type
> > sounds.
> 
> It's so tempting, and I think you/we should resist.  Your proposal 
> already has plenty in it.  Once we have a little bit of piano roll, 
> we'll want more and more and more features.  It's smarter for the 
> Audacity team to have a well thought through plan for bridging to 
> Rosegarden.  We don't, but we should.  We'll get a vastly more powerful 
> system by that route for similar work once all the extensions we'll want 
> to add to that tiny piano roll are taken into account!

Right, a piano roll may be a bit over the top. A simplification that
could be very useful and hardly require any extra work at all would be
to just have the ability to have LV2 synths generate chords instead of
single tones - all that would be needed would be a dropdown menu where
you could select between { "Single tone", "Minor", "Major",
"Diminished", "Augmented" } or something like that.

I'll update the proposal on the wiki again.


--ll