README.md

babble-projectrtp

Node module for projectrtp the RTP server. This, as well as projectrtp, is work in progress.

There is also a utility class for SDP manipulation which uses the sdp-transform library for parsing and generating.

Interface

Include and create an object to interface.

const projectrtp = require( "babble-projectrtp" )
const rtp = new projectrtp()

This will start a server listening. RTP servers then can connect to us so that we can distribute work to them.

Open a channel

A channel opens a stream between the ProjectRTP server and the UAC (or UAS).

rtp.channel( remote )
  .then( ch => {
    // Do something this the channel. Perhaps display info about it.
    console.log( ch )
  } )

Destroy a channel

  ch.destroy()
    .then( () => { console.log( "Channel destroyed" ) })

Functions

Mix

Mix 1 + n channels together.

Echo

A simple echo RTP back to other end.

  ch.echo()

Play (sound)

Play a sound 'soup' to the other end. See projectrtp for more information.

  ch.play( {
    "loop": true,
    "files": [
      { "wav": "ringing.wav", "loop": 6 },
      { "wav": "youare.wav" },
      { "wav": "first.wav" },
      { "wav": "inline.wav" }
    ]
  } )

Record

Can take options as per projectrtp: file = - filename to save as

In seconds up to MA max size (5 seconds?), default is 1 second RMS power is calculated from each packet then averaged using a moving average filter. poweraverageduration = moving average window to average power over

must have started for this to kick in - if left out will just start startabovepower =

When to finish - if it falls below this amount finishbelowpower =

used in conjunction with finishbelowpower - used in conjusnction with power thresholds i.e. power below finishbelowpower before this number of mS has passed minduration = < int > mSeconds

Must be above minduration and sets a limit on the size of recording maxduration = < int > mSeconds

  ch.record( {
    "file": "testfile.wav"
  } )

Or a more complex setup where speech is required to pass over to a processor (like speech recognition).

  ch.record( {
    "file": "testfile.wav",
    "startabovepower": 250,
    "finishbelowpower": 200,
    "minduration": 2000,
    "maxduration": 15000,
    "poweraverageduration": 1
  } )

Example

Example with using drachtio. See separate instructions for starting an RTP server.

const Srf = require( "drachtio-srf" )
const parseuri = require( "drachtio-srf" ).parseUri
const config = require( "config" )
const projectrtp = require( "babble-projectrtp" )

const srf = new Srf()
srf.connect( config.drachtio )


const rtp = new projectrtp()

srf.invite( ( req, res ) => {

  res.send( 180 )

  let parsedaor = parseuri( req.msg.uri )
  let remote = rtp.sdpgen( req.msg.body )

  // Grab a channel
  rtp.channel( remote )
    .then( ch => {

      let local = rtp.sdpgen()
                      .include( "pcmu pcma" )
                      .from( remote )
                      .setchannel( ch )

      const dlg = srf.createUAS( req, res, {
        localSdp: local.toString()
      } )

      .then( ( dlg ) => {

        switch( parsedaor.user ) {
          case "3":
            ch.play( { "loop": true, "files": [ { "wav": "uksounds.wav" } ] } )
            break
          default:
            ch.echo()
        }

        dlg.on( "destroy", () => {

          ch.destroy()
            .then( () => { console.log( "Channel destroyed" ) } )

          console.log( "Call ended" )
        } )
      } )
      .catch( ( err ) => {
        console.log( err )
      } )
  } )
} )

TODO

• More intelligent routing to pass calls off to multiple RTP servers for scalability. It only currently supports single server.
• Mix more than 2 channels (conference).
• Support video (pass through)
• Lots!

Refereces

• An Offer/Answer Model with the Session Description Protocol (SDP) RFC 3264
• SDP: Session Description Protocol RFC 4566