Inbound audio as ffmpeg stream

[oischinger]

I'm trying to implement 2-way audio for Eufy's Doorbells which support only P2P.
For that I would need to route the webrtc audio stream to an ffmpeg process that trancodes the audio and sends it to a tcp socket.

Is something like that possible with go2rtx?

[AlexxIT]

Well. First. How do you plan to get video and audio from P2P?

[oischinger]

I have a little process that provides audio and Video from the camera via 2 socket inputs which I defined in the go2rtc yaml. Works like charm

[AlexxIT]

Do you have open example, how do you do it?

[oischinger]

This is the project:

It's an HA addon based on your docker Container plus a python process to deal with the eufy camera via a websocket:
https://github.com/oischinger/eufyp2pstream

[AlexxIT]

Well. WebRTC > go2rtc > ffmpeg > python > nodejs. Very complicated. Needs time to understand all of this :)

[oischinger]

Yes, the stack is a bit crazy. The stack needs some work ;)
So far this is just a proof of concept. A/V worked fine. But I'm not sure if a backchannel is possible the way I think about it. Since my SDP from ffmpeg doesn't include the backchannel information I wonder if it's possible to establish a stream for it at all

[AlexxIT]

I thinks audio back to ffmpeg is possible. But not a priority in near future. Have other nice things to do.

[AlexxIT]

Exec source now support backchannel
https://github.com/AlexxIT/go2rtc/releases/tag/v1.9.0