/alexa-gpt

A tutorial on how to use ChatGPT in Alexa

Primary LanguagePythonMIT LicenseMIT

Alexa GPT

License: MIT

Boost your Alexa by making it respond as ChatGPT.

This repository contains a tutorial on how to create a simple Alexa skill that uses the OpenAI API to generate responses from the ChatGPT model.

Prerequisites

Step-by-step tutorial

1.

Log in to your Amazon Developer account and navigate to the Alexa Developer Console.

2.

Click on "Create Skill" and name the skill "Chat". Choose the primary locale according to your language.

name your skill

3.

Choose "Other" and "Custom" for the model.

type of experience

choose a model

4.

Choose "Alexa-hosted (Python)" for the backend resources.

hosting services

5.

You now have two options:

Or if you want to import the skill manually

  • Select "Start from Scratch", click "Create Skill" and go to step 6

template

6.

In the "Build" section, navigate to the "JSON Editor" tab.

7.

Replace the existing JSON content with the provided JSON content:

{
    "interactionModel": {
        "languageModel": {
            "invocationName": "chat",
            "intents": [
                {
                    "name": "GptQueryIntent",
                    "slots": [
                        {
                            "name": "query",
                            "type": "AMAZON.Person"
                        }
                    ],
                    "samples": [
                        "{query}"
                    ]
                },
                {
                    "name": "AMAZON.CancelIntent",
                    "samples": []
                },
                {
                    "name": "AMAZON.HelpIntent",
                    "samples": []
                },
                {
                    "name": "AMAZON.StopIntent",
                    "samples": []
                },
                {
                    "name": "AMAZON.NavigateHomeIntent",
                    "samples": []
                }
            ],
            "types": []
        }
    }
}

json_editor

8.

Save the model and click on "Build Model".

9.

Go to "Code" section and add "openai" to requirements.txt. Your requirements.txt should look like this:

ask-sdk-core==1.11.0
boto3==1.9.216
openai==1.3.3

10.

Create an OpenAI API key by signing up and clicking in "+ Create new secret key" in the API keys page.

11.

Replace your lambda_functions.py file with the provided lambda_function.py.

import logging
import ask_sdk_core.utils as ask_utils
from openai import OpenAI
from ask_sdk_core.skill_builder import SkillBuilder
from ask_sdk_core.dispatch_components import AbstractRequestHandler
from ask_sdk_core.dispatch_components import AbstractExceptionHandler
from ask_sdk_core.handler_input import HandlerInput
from ask_sdk_model import Response

# Set your OpenAI API key
client = OpenAI(
    api_key="YOUR_API_KEY"
)

logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)

class LaunchRequestHandler(AbstractRequestHandler):
    """Handler for Skill Launch."""
    def can_handle(self, handler_input):
        # type: (HandlerInput) -> bool

        return ask_utils.is_request_type("LaunchRequest")(handler_input)

    def handle(self, handler_input):
        # type: (HandlerInput) -> Response
        speak_output = "Chat G.P.T. mode activated"

        session_attr = handler_input.attributes_manager.session_attributes
        session_attr["chat_history"] = []

        return (
            handler_input.response_builder
                .speak(speak_output)
                .ask(speak_output)
                .response
        )

class GptQueryIntentHandler(AbstractRequestHandler):
    """Handler for Gpt Query Intent."""
    def can_handle(self, handler_input):
        # type: (HandlerInput) -> bool
        return ask_utils.is_intent_name("GptQueryIntent")(handler_input)

    def handle(self, handler_input):
        # type: (HandlerInput) -> Response
        query = handler_input.request_envelope.request.intent.slots["query"].value

        session_attr = handler_input.attributes_manager.session_attributes
        chat_history = session_attr["chat_history"]
        response = generate_gpt_response(chat_history, query)
        session_attr["chat_history"].append((query, response))

        return (
                handler_input.response_builder
                    .speak(response)
                    .ask("Any other questions?")
                    .response
            )


class CatchAllExceptionHandler(AbstractExceptionHandler):
    """Generic error handling to capture any syntax or routing errors."""
    def can_handle(self, handler_input, exception):
        # type: (HandlerInput, Exception) -> bool
        return True

    def handle(self, handler_input, exception):
        # type: (HandlerInput, Exception) -> Response
        logger.error(exception, exc_info=True)

        speak_output = "Sorry, I had trouble doing what you asked. Please try again."

        return (
            handler_input.response_builder
                .speak(speak_output)
                .ask(speak_output)
                .response
        )

class CancelOrStopIntentHandler(AbstractRequestHandler):
    """Single handler for Cancel and Stop Intent."""
    def can_handle(self, handler_input):
        # type: (HandlerInput) -> bool
        return (ask_utils.is_intent_name("AMAZON.CancelIntent")(handler_input) or
                ask_utils.is_intent_name("AMAZON.StopIntent")(handler_input))

    def handle(self, handler_input):
        # type: (HandlerInput) -> Response
        speak_output = "Leaving Chat G.P.T. mode"

        return (
            handler_input.response_builder
                .speak(speak_output)
                .response
        )

def generate_gpt_response(chat_history, new_question):
    try:
        messages = [{"role": "system", "content": "You are a helpful assistant."}]
        for question, answer in chat_history[-10:]:
            messages.append({"role": "user", "content": question})
            messages.append({"role": "assistant", "content": answer})
        messages.append({"role": "user", "content": new_question})
        response = client.chat.completions.create(
            model="gpt-3.5-turbo-1106",
            messages=messages,
            max_tokens=300,
            n=1,
            temperature=0.5
        )
        return response.choices[0].message.content
    except Exception as e:
        return f"Error generating response: {str(e)}"

sb = SkillBuilder()

sb.add_request_handler(LaunchRequestHandler())
sb.add_request_handler(GptQueryIntentHandler())
sb.add_request_handler(CancelOrStopIntentHandler())
sb.add_exception_handler(CatchAllExceptionHandler())

lambda_handler = sb.lambda_handler()

12.

Put your OpenAI API key that you got from your OpenAI account

open_api_key

13.

Save and deploy. Go to "Test" section and enable "Skill testing" in "Development".

development_enabled

14.

You are now ready to use your Alexa in ChatGPT mode. You should see results like this:

test

Please note that running this skill will incur costs for using both AWS Lambda and the OpenAI API. Make sure you understand the pricing structure and monitor your usage to avoid unexpected charges.