What is app_audiofork

app_audiofork helps you send audio from Asterisk to a websocket server easily.

For instance:

ASTERISK -> AUDIO STREAM -> WS APP SERVER

The primary objective of this module is to offer a straightforward integration with the Asterisk audiohooks API, providing an interface for developers to better analyze audio and run other processing tasks on it.

How to install it

You can use the Makefile to easily install app_audiofork.

To get started, please run:

make
make install

Load module

Afterwards, you can load the module with the following command:

asterisk -rx 'module load app_audiofork.so'

Configuring in dial plans

Here is a simple example of how to use "AudioFork()"

exten => _X.,1,Answer()
exten => _X.,n,Verbose(starting audio fork)
exten => _X.,n,AudioFork(ws://localhost:8080/)
exten => _X.,n,Verbose(audio fork was started continuing call..)
exten => _X.,n,Playback(hello-world)
exten => _X.,n,Hangup()

Configuring a Websocket server

In order to integrate the module, it is advised that you use a websocket server that is compliant with the latest standard. In other words, it should reliably send both text and binary data frames.

Here is a recommended Node.js Websocket module: WebSocket nodejs server

Example: simple integration

Below is an example that receives audio frames from the AudioFork app and stores them in a audio file.

const WebSocket = require('ws');

const wss = new WebSocket.Server({ port: 8080 });
var fs = require('fs');
var wstream = fs.createWriteStream('audio.raw');

wss.on('connection', function connection(ws) {
  console.log("got connection ");

  ws.on('message', function incoming(message) {
    console.log('received frame..');
    wstream.write(message);
  });
});

Converting raw audio to WAV

You can quickly convert any raw audio data to a another format, such as WAV, using the Sox command line tool.

For example:

sox -r 8000 -e signed-integer -b 16 audio.raw audio.wav

Sending separate Websocket streams

In a production scenario, it is common to handle both the incoming and outgoing legs of a call. The basic example doesn't do that, but it is certainly possible.

Below is an example including a WebSocket server that handles two connections and stores each stream in its a unique file.

Updated dialplan

[main-out]
exten => _.,1,Verbose(call was placed..)
same => n,Answer()
same => n,AudioFork(ws://localhost:8080/out,D(out))
same => n,Dial(SIP/1001,60,gM(in))
same => n,Hangup()

[macro-in]
exten => _.,1,Verbose(macro-in called)
same => n,AudioFork(ws://localhost:8080/in,D(out))

Node.js server implementation

const http = require('http');
const WebSocket = require('ws');
const url = require('url');
const fs = require('fs');

const server = http.createServer();
const wss1 = new WebSocket.Server({ noServer: true });
const wss2 = new WebSocket.Server({ noServer: true });
var outstream = fs.createWriteStream('out.raw');
var instream = fs.createWriteStream('in.raw');


wss1.on('connection', function connection(ws) {
  // ...
  console.log("got out connection ");

  ws.on('message', function incoming(message) {
    console.log('received out frame..');
    outstream.write(message);
  });
});

wss2.on('connection', function connection(ws) {
  // ...
  console.log("got in connection ");

  ws.on('message', function incoming(message) {
    console.log('received in frame..');
    instream.write(message);
  });

});

server.on('upgrade', function upgrade(request, socket, head) {
  const pathname = url.parse(request.url).pathname;

  if (pathname === '/out') {
    wss1.handleUpgrade(request, socket, head, function done(ws) {
      wss1.emit('connection', ws, request);
    });
  } else if (pathname === '/in') {
    wss2.handleUpgrade(request, socket, head, function done(ws) {
      wss2.emit('connection', ws, request);
    });
  } else {
    socket.destroy();
  }
});

server.listen(8080);

Live transcription demos

You can refer to the following demos for more complete integrations.

• Google cloud speech Transcribe Demo
• Azure demo integration
• Amazon transcribe demo

TLS support

AudioFork currently supports secure websocket connections. In order to create a secure websocket connection, you must add the "T" option to the app options.

For example:

AudioFork(wss://example.org/in,D(out)T(on))

Reconnecting closed sockets

It is also possible to setup basic backoff for reconnection. By default, Audiofork is configured to reconnect to the WS server, and after a preconfigured number of attempts it will close the connection. These parameters, however, can be adjusted.

To adjust the reconnection parameters, you can use the following parameters:

R(timeout_for_connection)
r(number of times to attempt reconnection)

For instance, the following example will set the reconnection timeout to 10 seconds and will attempt to reconnect five times.

AudioFork(wss://example.org/in,R(10)r(5))

Start an audio stream on demand

It is possible to start an audio stream for a live call. We can do this by using AMI (asterisk manager interface).

In essence, we can start start streaming audio for any active channel.

For a full example you can refer to the following NodeJS reference code. It uses the NodeJS-AsteriskManager library

Example: connect to AMI and call Audiofork

var ami = new require('asterisk-manager')('port','host','username','password', true);
ami.keepConnected();
ami.action({
  'action':' 'AudioFork',
  'channel':'PJSIP/myphone',
  'ActionID': '1234',
  'WsServer': 'ws://your_ws_server_url',
  'Options': 'D(both)',
  'Command': 'StartAudioFork'
}, function(err, res) {
  if (err) {
    console.error( err )
    return;
  }
  console.log(res)
});

Project roadmap

At this time, AudioFork is largely incomplete and has many updates planned.

The following updates are scheduled for the upcoming releases:

• Test and ensure module is fully compatible with Asterisk manager
• Store any WebSocket responses in a dialplan variable

Contact info

For any queries, please contact me directly:

Nadir Hamid <[email protected]>