openvrml-develop

[openvrml-develop] Sound in OpenVRML

[-oli- <ol...@fr...>]

Hi!

I've got some questions concerning sound in OpenVRML:

a) Why was sound support taken out of the current release?
b) Under which OS'es did the old sound support (sound.cpp in CVS) work?

Thanks for help!

	-oli-
	(Oliver Baum)

Re: [openvrml-develop] Sound in OpenVRML

[Braden M. <br...@en...>]

On Tue, 2002-03-12 at 06:09, -oli- wrote:
> Hi!
> 
> I've got some questions concerning sound in OpenVRML:
> 
> a) Why was sound support taken out of the current release?

It was disabled a few releases ago. It was a matter of convenience then.
And I intended for it to be replaced with something better. I still do;
but I didn't anticipate having to wait this long. In retrospect, I wish
I hadn't disabled it. However, I now have higher priorities than turning
it back on.

> b) Under which OS'es did the old sound support (sound.cpp in CVS) work?

Just Linux, I think.

If you want to submit a patch to enable the sound support, it would be
welcome.

In terms of something better, I want to do streaming. Since streaming
media APIs differ across platforms, we need to have an abstraction layer
analogous to what we have for the 3D renderer. Prerequisite for
streaming is asynchronous IO, which is something I'll be working on for
0.13.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Sound in OpenVRML

[-oli- <ol...@fr...>]

At 09:06 12.03.2002 -0500, you wrote:
>It [sound] was disabled a few releases ago. It was a matter of convenience 
>then.
>And I intended for it to be replaced with something better. I still do;
>but I didn't anticipate having to wait this long. In retrospect, I wish
>I hadn't disabled it. However, I now have higher priorities than turning
>it back on.

Oh, I understand!
Was there distance attenuation, spatalizing etc.?

>If you want to submit a patch to enable the sound support, it would be
>welcome.

I'm thinking about integrating OpenAL-support in OpenVRML, but there are 
major incompatibilities between the directivity model used in OpenAL 
(cones) and VRML (ellipsoids). I'm trying to find a solution here...

The problem also is that do not know how to generate a patch.
Do I need diff or something like this?

>In terms of something better, I want to do streaming.

With OpenAL this is possible - but not trivial.
I have also not the understanding of OpenVRML to see in which (main-?)loop 
the next chunk of the wave-file can be handed to the sound buffer...
Tips are welcome!

         -oli-
         (Oliver Baum)

Re: [openvrml-develop] Sound in OpenVRML

[Braden M. <br...@en...>]

Quoting -oli- <ol...@fr...>:

> At 09:06 12.03.2002 -0500, you wrote:
> >It [sound] was disabled a few releases ago. It was a matter of convenience
> 
> >then.
> >And I intended for it to be replaced with something better. I still do;
> >but I didn't anticipate having to wait this long. In retrospect, I wish
> >I hadn't disabled it. However, I now have higher priorities than turning
> >it back on.
> 
> Oh, I understand!
> Was there distance attenuation, spatalizing etc.?

I'm not sure... All the code is still there in CVS.

> >If you want to submit a patch to enable the sound support, it would be
> >welcome.
> 
> I'm thinking about integrating OpenAL-support in OpenVRML, but there are 
> major incompatibilities between the directivity model used in OpenAL 
> (cones) and VRML (ellipsoids). I'm trying to find a solution here...

I know Tom Flynn has done some work to this end. Hopefully he'll chime in.

I'm of the opinion that the spatialization/directivity code ought to be part of
OpenVRML. That approach means that

 * we have control over our consistency with the VRML97 spec.
 * we don't require much intelligence from the underlying sound library,
   meaning we can be portable to different audio renderers without too much
   hassle.

> The problem also is that do not know how to generate a patch.
> Do I need diff or something like this?

If you have a CVS working tree, "cvs diff -u". Otherwise you'll want to use
plain ol' diff. (But you really should have a CVS working tree.)

> >In terms of something better, I want to do streaming.
> 
> With OpenAL this is possible - but not trivial.

I would like to use GStreamer--for streaming both sound and video. On Windows
this would be handled with DirectX, so we need a good abstraction API.

> I have also not the understanding of OpenVRML to see in which (main-?)loop 
> the next chunk of the wave-file can be handed to the sound buffer...
> Tips are welcome!

Check out the old sound code.

-- 
Braden McDaniel                                   e-mail: br...@en...
http://endoframe.com                              Jabber: br...@ja...

Re: [openvrml-develop] Sound in OpenVRML

[Thomas F. <tf...@us...>]

Braden McDaniel wrote:

>Quoting -oli- <ol...@fr...>:
>
>>At 09:06 12.03.2002 -0500, you wrote:
>>
>>>It [sound] was disabled a few releases ago. It was a matter of convenience
>>>
>>>then.
>>>And I intended for it to be replaced with something better. I still do;
>>>but I didn't anticipate having to wait this long. In retrospect, I wish
>>>I hadn't disabled it. However, I now have higher priorities than turning
>>>it back on.
>>>
>>Oh, I understand!
>>Was there distance attenuation, spatalizing etc.?
>>
>
>I'm not sure... All the code is still there in CVS.
>

I'm about 99% sure there's no distance attenuation, spatializing or 
anything else except for looping support...the code only works on Linux 
and I think the only things that need to be done to get it to compile is 
to define HAVE_SOUND and add the sound code back to the makefiles. If I 
can find the time, I'll generate a patch to get this readded to the 
build until we have a better solution...

>>>If you want to submit a patch to enable the sound support, it would be
>>>welcome.
>>>
>>I'm thinking about integrating OpenAL-support in OpenVRML, but there are 
>>major incompatibilities between the directivity model used in OpenAL 
>>(cones) and VRML (ellipsoids). I'm trying to find a solution here...
>>
>
>I know Tom Flynn has done some work to this end. Hopefully he'll chime in.
>

-oli- contacted me a month or so ago about this and I wasn't able to 
give him any answers as I never made it that deep into integrating OpenAL...


<snip>

>>I have also not the understanding of OpenVRML to see in which (main-?)loop 
>>the next chunk of the wave-file can be handed to the sound buffer...
>>Tips are welcome!
>>
>
>Check out the old sound code.
>

NodeAudioClip::update (in vrml97node.cpp) is called on each loop so 
that's one place where you could do it...

Tom