Multiple channels

[joshua22s]

Is it possible to run multiple seperate channels? Where each channel has it own audio in- and output. That way I can have multiple calls running simultaneously.

[kingster]

Most normal humans are able to have conversation with 1 other party at a time.

Could you explain the scenario you have in hand that you are trying to solve by running multiple calls parallely each with their own audio in/put?

[joshua22s]

I work in broadcasting, primarily in radio. My goal is to set up a dedicated machine capable of handling multiple SIP channels simultaneously. Each channel should be routed to a separate input and output on my audio mixer. This would allow me to assign every call (or account) to its own channel strip on the mixer, enabling individual control over each connection.

[kingster]

What kind of audio hardware does this machine has? Because if it has to emit out that many channels to your mixer, then it has to have so many outputs channels.

Alternatively, why not use the hold facility? The reason is when you mute someone though a mixer they would hear only silence vs when they are put on mute they would hear the hold music (I have limited knowledge of the mixer in radio industry and its capabilities)

[joshua22s]

I will probably use a Dante virtual soundcard. With that I can make up to 64 inputs and 64 outputs (but i probably use 4 in 4 out max). Yes muting could work if I only use one channel at a time. Most of the times we use multiple channels at a time. We use sip connections a lot for intercom purposes.

[kingster]

It could be done (since we identify by deviceId), but I am not sure how someone who doesn't have access to those tools would test/build.

[joshua22s]

Is it possible to link the deviceIds to an account? For example when I do a login POST for a certain account that I also define the in- and output deviceIds when that specific account uses for making calls?

[kingster]

The feature doesn't exist on the current version, however, technically should be possible, since during the answer step there is an option to choose a specific device.

tinyphone/tinyphone/call.cpp

Lines 71 to 86 in 092fdd2

	AudioMedia aud_med;
	try {
	// Get the first audio media // for PJSIP version > 2.8
	aud_med = getAudioMedia(-1);
	} catch(...) {
	PJ_LOG(3, (__FILENAME__, "ERROR: Call [%d] OnCallMediaState Media Not Found", ci.id));
	return;
	}
	// Connect the call audio media to sound device
	AudDevManager& mgr = Endpoint::instance().audDevManager();
	PJ_LOG(3, (__FILENAME__, "Connecting Call [%d] to Media Device Input #%d , Output # %d", ci.id, mgr.getCaptureDev(), mgr.getPlaybackDev()));
	// This will connect the sound device/mic to the call audio media
	mgr.getCaptureDevMedia().startTransmit(aud_med);
	// And this will connect the call audio media to the sound device/speaker
	aud_med.startTransmit(mgr.getPlaybackDevMedia());

[joshua22s]

Alright, that looks promising. I don't have any experience developing in C++ but I will see if I can find some time to dive into it and write a PR.