Just a clarification. This is a lab experiment just for fun, YOU SHOULD NOT USE THIS IN PRODUCTION ENVIRONMENTS!!!! I have barely tested this library and it is highly likely to have some bad bugs. Use this library at your own risk... or take the time to review it, fix it and improve it before any serious use.

RTCVoiceChat

This is just an experiment about WebRTC and audio channels. I looked for a simple approach. First I made a base class (P2PVoiceChannel) to manage one channel to send audio from source (named master role) to destination (named slave role). Then I made a container class (P2PVoiceChat) that instance 2 times the P2PVoiceChannel, one as EMITTER and other for RECEIVER.

To create a simple 2 agents voice chat, first you need to instantiate P2PVoiceChat at each end, then set for each instance a signal sender function, and finally leave it up to the instance to handle the abstract event onsignalreceived (you need to implement your own signal sender and receiver).

sequenceDiagram
  title Connection process
  participant E1 as AGENT 1 - Emitter
  participant R2 as AGENT 2 - Receiver

  E1->>R2: Send INVITE
  R2--xE1: Send REJECT
  Note over E1,R2: Call rejected, end of transaction
  R2->>E1: Send ACCEPT
  activate E1
  Note left of E1: Create<br/>OFFER
  E1->>R2: Send OFFER
  deactivate E1
  activate R2
  Note right of R2: Create<br/>ANSWER
  R2->>E1: Send ANSWER
  deactivate R2
  E1->>R2: Send READY
  Note over E1,R2: Channel between AGENT 1 Emitter<br/>and AGENT 2 Receiver established
  %% Emitter of agent 1 now has an open channel to Receiver of agent 2
  create participant E2 as AGENT 2 - Emitter
  R2->>E2: Pass control to his emitter
  create participant R1 as AGENT 1 - Receiver
  E2->>R1: Send INVITE
  R1->>E2: Send ACCEPT
  activate E2
  Note left of E2: Create<br/>OFFER
  E2->>R1: Send OFFER
  deactivate E2
  activate R1
  Note right of R1: Create<br/>ANSWER
  R1->>E2: Send ANSWER
  deactivate R1
  E2->>R1: Send READY
  Note over E2,R1: Channel between AGENT 2 Emitter<br/>and AGENT 1 Receiver established
  Note over E1,R1: Now 2 channel by agent are open and bidireccional communication is established
  
%%{init:{'themeCSS':':is(#root-4,#root-7) rect,#actor4,#actor7 { fill: #CAFFBF; stroke: #8EC383; };:is(#root-5,#root-6) rect,#actor5,#actor6 { fill: #A0C4FF; stroke: #6589C3; };g:has(.actor.actor-bottom) { display: none; }'}}%%
    

Loading

How to use

First you need instantiate RTCVoiceChat:

const voiceChat = new P2PVoiceChat({
    channelId: 'testvchat', // any text 
    signalSend: send // the function name that send signal to the other end
});

signalSend pass 4 parameters to the function: channelId, type, data and src (src carries the role of the sender, master or slave).

Then, assuming that you have a signal receiver in a guessed object signaling, you must leave it up manage signals to voiceChat instance:

// Let's assume that e carries the data we need to send to the voiceChat signal processor.
signaling.onmessage = e => voiceChat.processSignal(e.type, e.data, e.src);

The e parameter received in onmessage should have at least type, data and src.

P2PVoiceChat constructor parameters

const voiceChat = new P2PVoiceChat({
    channelId: <text>            // Any text 
    signalSend: <function>       // Function that send signal to the other end,
    iceServers: <object>         // iceServers data. 
                                 // Default is: { iceServers: [{ urls: ["stun:stun.gmx.net"] }] }
    enableSounds: <boolean>      // Enable or not calling and ringing tone at each end. Default true.
    audiometer: <boolean>        // If true send to onaudiometer function audio intensity (0 to 100);
    onaudiometer: <function>     // Function that allow to draw a VUMeter or something like that.
                                 // P2PVoiceChat pass 2 parameters: intensity (1 to 100) and P2PVoiceChat instance
    onend: <function>            // It is executed once the communication is terminated when one of the 2 ends hangs up the call.
                                 // P2PVoiceChat pass as parameter P2PVoiceChat instance
    onreceiveinvite: <function>  // It is executed when the receiver part of P2PVoiceChat receive an "invite".
    onreceiveaccept: <function>  // It is executed when the emitter part of P2PVoiceChat receive an "accept".
    onreceivereject: <function>  // It is executed when the emitter part of P2PVoiceChat receive a "reject".
    onvoicechatready: <function> // It is executed when both channels of P2PVoiceChat (emitter and receiver) are ready to talk.
});

P2PVoiceChat useful instance methods

P2PVoiceChat.call()    // start the call sending an invite to the other end. 
P2PVoiceChat.reject()  // reject the call after the other end send an **invite**. 
P2PVoiceChat.accept()  // accept the call after the other end send an **invite**.
P2PVoiceChat.end()     // end the call and send an **end** signal to the other end. 

P2PVoiceChat useful instance properties

P2PVoiceChat.incurse   // true if the call is in curse.

Yes, that's all!! Take a look to the code if you need some more "hacky" access.

How to test

Clone the repository and load index.html in 2 differents tabs of the browser. To pass signals this test use BroadcastChannel.