Hi Thomas, Sorry for the late reply. If I read your post correctly, I think you're modeling the solution based on the assumption that we can't use multiple sound devices with the conference bridge. That's not true actually. You can create another sound device (assuming PA Jack supports this), create and connect a splitcomb to the snd dev, then create a rev_port from the splitcomb, and register this rev_port to the bridge (just as you register any other port to the bridge). There is a snippet for this in pjsua_app.c, see the stereo_demo() function towards the bottom of the file. Alternatively, since you don't need mixing, why don't you just disable the bridge altogether, and connect the stream directly to the corresponding jack sound dev? This is explained here: http://trac.pjsip.org/repos/wiki/FAQ#pjsua-lib-perf cheers Benny On Wed, Jan 21, 2009 at 10:28 AM, Thomas Kluge <T.Kluge at gmx.com> wrote: > Hi all, > > We try to build a VoIP application for simulating several audio media > features on top of the pjsip stack. This application should have the > following features related to the pjsip-stack: > > - support multiple ports of a jack client > - connection between transport streams and ports of a jack client (no > mixing) > > In my opinion there are two different approaches to a solution, every one > leads to several poblems I want to describe in relation to the model used. > I guess, that a solution would require significant changes/extensions > within the pjsip stack. > > *** Recording model (suitable in both models): *** > > Jack client x input port - conferenc bridge master port - tx to all > transport streams (that's already the case) > > *** Playback models: *** > > Model 1 > ====================================================================== > > - use 1 jack input client and connect it to the conferenc bridge master > port > - use n-1 jack clients with 1 output port each and register them to > different ports at the conference bridge. > - That would require a support of n sounddevices simultaneously (todo). > > Jack client 1 output port - conferenc bridge - rx transport stream 1 > Jack client 2 output port - conferenc bridge - rx transport stream 2 > Jack client 3 output port - conferenc bridge - rx transport stream 3 > ... > Jack client n-1 output port - conferenc bridge - rx transport stream n-1 > > Issues: > > - As I see there is currently only one sounddevice supported by the > conderence bridge. This sounddevice has been selected within collected > device information gathered from hostapi/portaudio. > - Setting a new device will clear !! the previous one and register the new > one to the master conference port. > > Solution: > > - change the function pjsua_set_snd_dev() and related functions to handle > more than one sounddevice at once > - mirror the master port creation to other conference ports. (I don't know > if this is possible using the same get_frame/put_frame callbacks of the > master port.) > - assign every sounddevice a separate conference port > - Ensure that sounddevce functions like OpenStream() can be called on > different sounddevices simultaneously > > > Model 2 > ====================================================================== > > - use a jack client with one input port and n-1 output ports and connect > it to the conference bridge master port. > - That would require support of multiple channels from PortAudio to > conference bridge (todo) > > > Jack client x output port 1 - conferenc bridge - rx transport stream 1 > Jack client x output port 2 - conferenc bridge - rx transport stream 2 > Jack client x output port 3 - conferenc bridge - rx transport stream 3 > ... > Jack client x output port n-1 - conferenc bridge - rx transport stream n-1 > > Issues: > > Starting from Jack I retraced the output port/channel information to the > conference bridge. I figured out that the information about multiple > ports/channels of a sounddevice gets lost in the transition from portaudio > (functions NonAdaptingProcess/AdaptingOutputProcess) to internal > functions like PaPlayerCallback. It looks like only the outputbuffers of > the > first port/channel will be processed within pjspip . (variable void > *output, > array which holds per channel output-buffers on the PortAudio-side) > > Solution: > > - provide, handle the channel / outputbuffer information from PortAudio to > internal processing structures > - disable mixing in conference bridge > - connect output ports with transport streams (function get_buffer()) > (I have no idea how I can resolve this using the current structures). > > > Any thoughts about that, did anybody something similar before? > > -- > Thanks And Regards, > Thomas Kluge > > _______________________________________________ > Visit our blog: http://blog.pjsip.org > > pjsip mailing list > pjsip at lists.pjsip.org > http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20090121/72ffb0a9/attachment.html>
