audio asterisk video other mixer

 https://groups.google.com/forum/#!msg/discuss-webrtc/B3iDAJ2XtVI/lsoEfmuJNFUJ

Dear all,

we've just published a public demo for you to access. This reflects the exciting work to add WebRTC support to our platform as I've described it here:

     https://groups.google.com/forum/?fromgroups#!topic/discuss-webrtc/YhI5fUpx1jc

The demo is a fully functional Meetecho room with WebRTC access to the audio/video, and includes the same set of features we usually make available when providing remote participation support to IETF meeting sessions. As such, it should be a nice preview of what you might get attending the sessions we'll make available in the upcoming IETF meeting in Vancouver ( http://www.ietf.org/meeting/84/index.html), especially with respect to the WebRTC support we just added.

The application creates a single PeerConnection object with the conferencing system: both the audio and video streams are mixed, meaning that no matter how many active participants are in the room, you'll always get a single stream back. To do so, we added VP8 support to our videomixer, and support to the RTCP FIR packet in order to request a key frame as soon as a user is allowed to contribute video to the mix. The other audio/video points of access are active in the room as well, so WebRTC, RTP applet and SIP users will be able to interact with each other, and passive users (RTSP/RTMP) will get everything as well. All those points of entry are bridged together.

If you want to give it a try, just click on this link:

     http://srv137.conf.meetecho.com/WebLite/login.jsp?event=webrtc

and follow the instructions. This is a public room and so everyone is free to play with it. That said, this is one of the servers we use for development so the session might not always be up and running: in case this happens, just try again after a few minutes.

A few hints to get it working:

• The audio/video part is not started by default when you join the room: you can start it in the audio/video tab, choosing the WebRTC icon to try the WebRTC access to the call. The application should autodetect support for the MediaStream and PeerConnection objects: as far as I know, it should work fine for every release of Chrome after 1180, with the exception of those few builds that I read didn't work on Windows (~1200), but I didn't test them all.
• A successfull call will bring up a floating window, showing the remote video (a static logo, if no participant is active) and a few controls. We've noticed that the remote stream doesn't seem to always immediately appear after a call: if so, wait a few seconds and it should be fine.
• When you're in the call, you're muted by default, since both audio and video are moderated by BFCP: to contribute your audio and/or your video to the mix, you need to send a floor request (the 'I want to talk', 'I want to be seen' buttons). Since the video is mixed, by default your own stream is not displayed in the mix you get back: if, after a successfull floor request, you want to appear in your own mix anyway (e.g., to test the videomixing stuff), press the mirror icon that will appear.
  

For all the other features, there's a nice tutorial here if you want to know more:

     http://ietf84.conf.meetecho.com/index.php/Web_Client

We're really excited about WebRTC, so we'd love some feedback on what we did to support it. If you have any, feel free to contact us, either here or privately (team [at]  meetecho.com) with any feedback or suggestion you might have for us!

Lorenzo

Hi Pablo,

thanks for your support, it's really appreciated!

The video quality at the moment is not perfect not because of UDP vs TCP (UDP is actually usually the best choice to carry realtime RTP packets) but because of our current videomixer implementation. We mix all video streams as a per-user QCIF video, right now, and since Chrome currently only sends 640x480 streams, the videomixer has sometimes a hard time donwsizing and transcoding active users. There are a few issues with key frames as well, but hopefully we'll deal with them soon enough!

We use Asterisk, an open source PBX, server side to act as a central node for all audio/video participants: Asterisk takes care of mixing audio, while it delegates the video mixing and composition to an external videomixer component we developed ourselves. This already worked for SIP and PSTN users, so the tricky part in WebRTC so was to add support for VP8 and the related RTP packetization in there as well.

RTMP instead is only available as a passive stream, meaning you can't use it to also contribute to the mix: we basically get the current mix in the PBX and make it available as RTSP and RTMP streams too (audio only right now, but we'll add video there as well).

Lorenzo