Amazon Alexa integration

Introduction

Amazon Alexa is a cloud-connected voice assistant that can be easily programmed to send requests to remote services that perform a variety of tasks such as turning lights on and off, opening doors, reading news or predicting weather.

Alexa usually runs on Amazon Echo smart speakers, but it can also run on smartphones.

To invoke an external service with a voice command it is required to develop a Skill, that is an Alexa application that runs in the Amazon cloud.

In this project we will show how to develop a Skill that invokes actions of a Brain4it module through different voice commands.

Programming the Alexa Skill

To create an Alexa Skill follow the next steps:

Programming the Brain4it module

For each intent Alexa performs an HTTP POST request to our endpoint passing a JSON object that contains the intent name and the values of its slots:

{
  "version": "1.0",
  "session": {
    "new": true,
    "sessionId": "amzn1.echo-api.session.0ac9a463-bfe7-4192-9b72-30e3efd25373",
    "application": {
      "applicationId": "amzn1.ask.skill.7a543b88-f041-4a61-9d60-20a6622393bf"
    },
    "user": {
      "userId": "amzn1.ask.account.AHL5GWE6HXWGHTMLPWSX65WO..."
    }
  },
  "context": {
    "System": {
      "application": {
        "applicationId": "amzn1.ask.skill.7a543b88-f041-4a61-9d60-20a6622393bf"
      },
      "user": {
        "userId": "amzn1.ask.account.AHL5GWE6HXWGHTMLPWSX65W..."
      },
      "device": {
        "deviceId": "amzn1.ask.device.AFUZX7KVT6KT724...",
        "supportedInterfaces": {}
      },
      "apiEndpoint": "https://api.eu.amazonalexa.com",
      "apiAccessToken": "eyJ0eXAiOiJKV1QiLCJhb..."
    },
    "Viewport": {
      "experiences": [
        {
          "arcMinuteWidth": 246,
          "arcMinuteHeight": 144,
          "canRotate": false,
          "canResize": false
        }
      ],
      "shape": "RECTANGLE",
      "pixelWidth": 1024,
      "pixelHeight": 600,
      "dpi": 160,
      "currentPixelWidth": 1024,
      "currentPixelHeight": 600,
      "touch": [
        "SINGLE"
      ]
    }
  },
  "request": {
    "type": "IntentRequest",
    "requestId": "amzn1.echo-api.request.2ef418c7-517f-405a-84b4-163782a9beae",
    "timestamp": "2018-12-21T16:41:26Z",
    "locale": "en-US",
    "intent": {
      "name": "add",
      "confirmationStatus": "NONE",
      "slots": {
        "number_a": {
          "name": "number_a",
          "value": "3",
          "confirmationStatus": "NONE",
          "source": "USER"
        },
        "number_b": {
          "name": "number_b",
          "value": "4",
          "confirmationStatus": "NONE",
          "source": "USER"
        }
      }
    }
  }
}

The endpoint performs the corresponding action and returns a JSON object that may indicate a text to speech:

{
  "body": {
    "version": "1.0",
    "response": {
      "outputSpeech": {
        "type": "PlainText",
        "text": "The result of the operation is 7"
      }
    }
  }
}

So the external function that will process the Alexa requests could be like this:

(function (context json)
  (local
    response
    type
    locale
    text
    intent
    intent_func
  )
  (set message (parse json "format" => "json"))
  (set type message/request/type)
  (set locale message/request/locale)
  (if (not (has MESSAGES locale))
    (set locale "en-US")
  )
  (cond
    (when (= type "LaunchRequest")
      (set text
        (get (get MESSAGES locale) "ready")
      )
    )
    (when (= type "IntentRequest")
      (set intent message/request/intent)
      (set intent_func (get intents intent/name))
      (set text (call intent_func intent locale))
    )
    (when true
      (set text
        (get (get MESSAGES locale) "unsupported")
      )
    )
  )
  (set context/response-headers
    ("content-type" => "application/json")
  )
  (set response
    (list
      "version" => "1.0"
      "response" =>
      (list
        "outputSpeech" =>
        (list "type" => "PlainText" "text" => text)
      )
    )
  )
  (string response "format" => "json")
)

For each intent type, a function with the intent name must be declared inside the intents list:

(
  "sum" =>
  (function (intent locale)
    (local number_a number_b)
    (set number_a
      (number intent/slots/number_a/value)
    )
    (set number_b
      (number intent/slots/number_b/value)
    )
    (concat
      (get (get MESSAGES locale) "sum")
      (+ number_a number_b)
    )
  )
  "open_door" =>
  (function (intent locale)
    (open_door)
    (get (get MESSAGES locale) "open_door")
  )
  ...
)

These functions return the text that Alexa will play on the smart speaker.

The sum function adds the values of the number_a and number_b slots and returns a message with the result of this operation.

If we say "Ask robot brain sum 3 plus 4" the sum function will be invoked and Alexa will respond: "The result of this operation is 7."

Download module
Top