Skip to content

Latest commit

 

History

History
102 lines (74 loc) · 2.71 KB

README.md

File metadata and controls

102 lines (74 loc) · 2.71 KB

MMM-Chat

MagicMirror Module to Chat with ChatGPT

Overview

This module integrates ChatGPT into your MagicMirror setup, allowing you to interact with an AI assistant directly through your mirror.

Key Features:

  • Customize the AI's personality.
  • Speech-to-text integration using Google STT or Whisper API.

Installation

  1. Clone the repository:

    git clone https://github.com/magicmirror-sdmy/MMM-Chat.git
cd MMM-Chat

  2. Install dependencies:

    pip3 install -r requirements.txt

  3. Add credentials to the MMM-Chat directory (see below for details).

Configuration

Default Installation Path

This module assumes the default installation path is /home/pi/MagicMirror. If you are using a different location, update node_helper.js with the correct directory.

Customizing the AI Personality

You can change the personality of the AI by editing line 26 in MMM-Chat.js. The default personality is set to "Drunk AI."

Example:

To make the AI sarcastic, use the following prompt:

"Your name is Marvin and you are a paranoid android that reluctantly answers questions with sarcastic responses."

Update the MMM-Chat.js file with the desired prompt.

Google Speech-to-Text Setup

Prerequisites

You will need:

  1. Google service account credentials.
  2. A billing account. Google offers a free quota; stay within this limit to avoid charges.

Steps

  1. Visit the Google Cloud Console.
  2. Create a new project or select an existing one.
  3. Navigate to APIs & Services > Credentials.
  4. Click Create credentials and select Service account key.
  5. Choose a service account name and assign the Editor role.
  6. Go to the service account and click Add key.
  7. Select JSON as the key type and click Create.

The JSON key file will be downloaded. Rename it to credentials.json and place it in the MMM-Chat directory.

Testing Speech-to-Text

To verify the setup, run the following command:

python3 transcript.py

Speak into the microphone, and the transcription of your words should appear.

Whisper API Integration

A script is provided for use with the Whisper API. However, in testing, Google STT offered better accuracy. If you'd like to use Whisper:

  1. Modify line 21 in node_helper.js to use whisper.py.
  2. Install the OpenAI Python library: 
    pip3 install openai

MagicMirror Configuration

Add the following lines to your config.js:

{
    module: "MMM-Chat",
    config: {}
},

Todo

  • Add a configuration option in config.js to support different AI personalities through prompts.