Skip to content

Latest commit

 

History

History
172 lines (125 loc) · 4.93 KB

README.md

File metadata and controls

172 lines (125 loc) · 4.93 KB
Raw

Matrix ChatGPT Voice Bot

Matrix ChatGPT Voice Bot is a Python GitHub repository that allows users to communicate with an AI chatbot using voice-to-text technologies powered by OpenAI and to generate images by Midjourney. It uses the GPT-3.5 Turbo model for generating text and engineering the prompts.

img

Supported commands

Unrestricted commands

  • help: Shows this help

Restricted commands

  • c: Chat with the chatGPT.
  • g: Generate images. Just type g [description] and the bot will generate the prompts and images.
  • clear: Clears the chat history.
  • openai: reset openai api key
  • replicate: reset replicate token

Requirements

  • apt-get install ffmpeg build-essential gcc libolm-dev
  • Python <= 3.10
  • simplematrixbotlib >= 2.8
  • pyyaml
  • openai
  • pydub
  • SpeechRecognition
  • replicate
  • Pillow

Getting started

Create a bot account

Then you need to create an account for the bot on the server, like you would do with any other account. A good username is chatgpt-voice-bot. If you want to use token based login, note the access token of the bot. One way to get the token is to log in as the bot and got to Settings -> Help & About -> Access Token in Element, however you mustn't log out or the token will be invalidated. As an alternative you can use the command

curl -X POST --header 'Content-Type: application/json' -d '{
    "identifier": { "type": "m.id.user", "user": "YourBotUsername" },
    "password": "YourBotPassword",
    "type": "m.login.password"
}' 'https://matrix.YOURDOMAIN/_matrix/client/r0/login'

Once you are finished you can start the installation of the bot.

Installation (manual)

The installation can easily be done via pip. It is recommended to use a virtual environment.

$ pip install .

Configuration

Configure the bot with a file named config/config.toml. It should look like this

[ simplematrixbotlib.config ]
  description = "This is a simple matrix bot library created by liuyang for audio recognition and text to AI models like (chatgpt | mdjurney)"

  HOMESERVER = "https://matrix.example.com"
  USERNAME = ""
  PASSWORD = ""
  ACCESS_TOKEN = ""
  OPEN_AI_KEY = ""
  ENABLE_ENCRYPTION = "true"
  REPLICATE_API_TOKEN = "" # key from replicate

It is also possible to use environment variables to configure the bot. The variable names are all upper or lower case.

Start the bot

Start the bot with

python -m matrix_chatgpt_voice_bot.main

and then open a Direct Message to the bot. The type one of the following commands.

Install using docker (recommended)

docker run -d --name matrix-chatgpt-voice-bot \
  -e HOMESERVER="https://synapse.example.com" \
  -e USERNAME="chatgpt-voice-bot" \
  -e PASSWORD="password" \
  -e ACCESS_TOKEN="" \
  -e OPEN_AI_KEY="" \
  -e REPLICATE_API_TOKEN="" \
  -e ENABLE_ENCRYPTION="true" \
  ly1998117/matrix-chatgpt-voice-bot:latest

If you use ACCESS_TOKEN you can leave PASSWORD empty. The OPEN_AI_KEY and REPLICATE_API_TOKEN is optional, you can leave it empty and reset them with the openai and replicate commands to the bot.

Install using docker-compose

To use this container via docker you can create the following docker-compose.yml and start the container with docker-compose up -d. Explanation on how to obtain the correct values of the configuration can be found in the Manual installation section.

version: "3.7"

services:
  matrix-chatgpt-voice-bot:
    image: ly1998117/matrix-chatgpt-voice-bot:latest
    environment:
      LOGGING_LEVEL: DEBUG
      HOMESERVER: "https://synapse.example.com"
      USERNAME: "chatgpt-voice-bot"
      PASSWORD: "password"
      ACCESS_TOKEN: ""
      ENABLE_ENCRYPTION: "true"
      OPEN_AI_KEY: ""
      REPLICATE_API_TOKEN: "" # key from replicate

Automatically (re-)start the bot with Systemd

To have the bot start automatically after reboots create the file /etc/systemd/system/matrix-chatgpt-voice-bot.service with the following content on your server. This assumes you use that you place your configuration in /matrix/config/config.yml and the bot in /matrix. If you use a different location, adjust the paths accordingly.

[Unit]
Description=matrix-chatgpt-voice-bot

[Service]
Type=simple

WorkingDirectory=/matrix
ExecStart=python3 -m matrix_chatgpt_voice_bot.main

Restart=always
RestartSec=30
SyslogIdentifier=matrix-chatgpt-voice-bot

[Install]
WantedBy=multi-user.target

After creating the service reload your daemon and start+enable the service.

$ sudo systemctl daemon-reload
$ sudo systemctl start matrix-chatgpt-voice-bot
$ sudo systemclt enable matrix-chatgpt-voice-bot

End-to-End Encryption

The bot supports E2E encryption now. This is a bit safer and also allows to create direct messages (which are by default encrypted). This will be enabled by default.

Related Projects