This package demonstrates how to use the ARI External Media feature to transcribe the audio from a bridge using the Google Speech APIs.
- A functional Asterisk 16.6.0+ installation.
- A conference bridge or phone configured.
- Node.JS version 10 or greater.
- Google Speech API credentials set in environment variable GOOGLE_APPLICATION_CREDENTIALS.
See https://cloud.google.com/speech-to-text/docs/ for more information.
Run npm install from the top of the source tree.
This will install the required npm packages including node-ari-client and @google-cloud/speech.
You can then run the transcriber as bin/ari-transcriber. If you add the -g
option to npm install to install system wide, you can just run ari-transcriber.
$ ari-transcriber --help
ari-transcriber [options] <dialstring>
Start the transcription server
Positionals:
dialstring Extension to dial such as "Local/1234"
Incoming Audio Server
--format, -f Asterisk format/codec [string] [choices: "ulaw", "slin16"] [default: "ulaw"]
--listenServer, -l Address and port on which to listen for audio from Asterisk [string] [default: "127.0.0.1:9999"]
Speech
--speechModel Google Speech API model [string] [choices: "phone_call", "video", "default"] [default: "default"]
--speechLang BCP-47 Language code. en-US, fr-CA, etc. [string] [choices: "en-US", "fr-CA"] [default: "en-US"]
--speakerDiarization Outputs words associated to speaker index to the console. [boolean] [default: false]
ARI
--ariServerUrl, -a The URL for the Asterisk instance [string] [default: "http://127.0.0.1:8088"]
--ariUser, -u The user configured in ari.conf [string] [default: "asterisk"]
--ariPassword, -p The password for the user configured in ari.conf [string] [default: "asterisk"]
Transcription WebSocket
--sslCert WebSocket secure server (wss) certificate. If omitted, a non-secures (ws) websocket server will be created [string]
--sslKey WebSocket secure server key. Must be supplied if sslCert is ssplied [string]
--wssPort WebSocket server port. If omitted, no websocket server will be started [number]
Options:
--version Show version number [boolean]
--help Show help [boolean]
--audioOutput, -o A file into which the raw audio from Asterisk can be written [string]
Examples:
ari-transcriber --format=slin16 --sslCert=/etc/letsencrypt/live/myserver/fullchain.pem
--sslKey=/etc/letsencrypt/live/myserver/privkey.pem --wssPort=39990 --speakerDiarization 'Local/1234'
The ari-transcriber performs several tasks:
- Creates an ari-client instance
- Creates a WebSocket server from which the live transcription can be accessed
- Starts an audio server to receive the audio from Asterisk
- Creates an instance of Google Speech Provider that takes the audio from the server, transcribes it, and sends the transcription out the websocket
- Uses the ARI instance to:
- Create a mixing bridge
- Create a Local channel which dials the conference bridge to be monitored
- Place the local channel into the mixing bridge
- Create an External Media channel that directs the audio to the audio server
- Place the External Media channel into the mixing bridge
Why the Local channel and bridge? To keep the sample as simple as possible, it's assumes that a conference bridge is already available. Since that bridge wouldn't be controlled by ARI/Stasis, we can't just add the External Media channel directly to it. Instead we have to create a Local channel that dials the conference bridge, then bridge that channel with the External Media channel. If you were ading External Media capabilities to your own application, chances are that your app would already control the participant bridge and you wouldn't need the Local channel and mixing bridge.
You don't need the WebSocket transcription server to try this. Just a phone to call.
$ export GOOGLE_APPLICATION_CREDENTIALS=<path to Google API credentials>
$ ari-transcriber --format=slin16 'Local/1234'