Mumble Client & Server SDK

[pwfran]

Hi,

we had conversation here

https://forums.mumble.info/topic/2209-mumble-client-amp-server-sdk/

but now it seems to be archived. I am trying to create my custom voice client using libmumble, based on example client in libmumble-master/examples/ExampleClient directory. Modifying this example client I am able to connect to server and receive voice packets from stock client on another machine, but I cannot decode them and do further processing.
Before the old forum was archived, @davidebeatrici asked for my code - I am attaching modified main.cpp file I use (named main.cpp.txt because it refused to attach .cpp). When I am talking to the remote client, custom client receives UDPTunnel messages (line 140). Then it passes

if (pack(msg))

condition (line 146) but fails to pass

if(msg.pack(audio))

(line 149).

So, I get sequence of printed messages:
=== feedback.pack - type: UDPTunnel
= UDPTunnel

UDPTunnel pack(msg)

Is this some basic mistake ? I would appreciate any support.

Thanks a lot for everyone for any help I got so far and sorry for being annoying - I couldn't find any example of using libmumble to create fully functional client and documentation is quite limited.

Best regards,
Pawel
main.cpp.txt

[davidebeatrici]

Hi,

Thank you for the recap and sorry about the archived thread, it seems to be missing our latest messages.

Could you check whether the function is returning here, please?

Thanks a lot for everyone for any help I got so far and sorry for being annoying - I couldn't find any example of using libmumble to create fully functional client and documentation is quite limited.

No worries, documentation is indeed missing (aside from the examples).

[pwfran]

It is sent from Linux, OpenSUSE 15.4 - mumble client from built in package (mumble-1.4.230-bp154.1.55.x86_64). Does standard Mumble client use the same format as libmumble or maybe not ?

[davidebeatrici]

Oh, sorry for taking so long to realize what the issue is... there is a new format for UDP packets in 1.5.x: mumble-voip/mumble#5594

Backward compatibility is provided in libmumble and the legacy format can be used fairly easily:

#include <mumble/PacketDataStream.hpp>

auto data = pack.data();
if (data.size() < 1 + 1 + 1) {
	// Audio packets must contain at least:
	//
	// - One header byte.
	// - One byte varint-encoding the sequence number.
	// - One byte of audio payload.
	return;
}

if (legacy::udp::type(data) != legacy::udp::Type::VoiceOpus) {
	return;
}

// const uint32_t targetOrContext = data[0] & 0x1f;

PacketDataStream stream(data.subspan(1));

uint32_t senderSession;
stream >> senderSession;

uint64_t frameNumber;
stream >> frameNumber;

uint64_t header;
stream >> header;

// const bool isLastFrame = header & 0x2000;

const auto opusData = stream.seek().first(header & 0x1fff);

[pwfran]

Thanks, I put this code inside:

else if(type == Type::UDPTunnel)
{
tcp::Message::UDPTunnel msg;
udp::Message::Audio audio;

        if (pack(msg)) 
        {
             // YOUR CODE HERE

       }
  }

And if (legacy::udp::type(data) != legacy::udp::Type::VoiceOpus) condition fails.
It seems that legacy::udp::type(data) is 0 while legacy::udp::Type::VoiceOpus is 4.

[davidebeatrici]

What's the packet size?

[pwfran]

When I say something to the connected peer, my compiled client gets a number of packets, data.size() is usually 105 - but sometimes there are single packets with other sizes: 55 or 104. But 105 dominates the set.