A prototype of the showcase to control OTT devices by voice commands.

General Workflow

• The OTT app open a WebSocket to a service after starting and authenticate with the user
• This service can handle voice commands by provide endpoints (e.g. for Dialogflow) for authenticated users
• If this service gets a voice command it send this command to all connected devices over the WebSockets
• The OTT app gets the command and can react on that

Developing / Testing

• The www/index.html opens the WebSocket
• The server.js provides GET /send/:intent to test the sending of commands
• The intent must match to a special video command (e.g. play) or an image filename in img/intents
• Different instances/users can be tested independently by providing an uuid
• www/index.html?uuid=test
• GET /send/:intent?uuid=test

Integration in Dialogflow

• The server.js provides POST /dialogflow/:uuid which must be defined in Dialogflow as fulfilment endpoint
• The Voice commands responses differently if responses are defined in the responses.json, otherwise with an Ok

Known issues

• On AWS WebSockets are not provided by default in Beanstalk apps
• If the server restarts all WebSocket connections are lost, the browser tab of the www/index.html must be refreshed