OpenAI Assistant

An AI assistant powered by OpenAI ChatGPT4

Note that this branch may not be stable please use releases

• Type to F.R.I.D.A.Y. and get typed responses
• Talk to F.R.I.D.A.Y. and get verbal responses

Known Bugs

• Calling ai whisper models on an M1 or M2 Mac CPU outputs an error every time

Next Patch

• Fully offline assistant
• Basic action functions (windows)
• Custom Model Training
• Linux & MacOS documentation
• Improved stabilty & permormance

Installation

1. First, clone the the repo to your machine. 2. Next, install Anaconda and run as administrator to install dependencies. 3. Then, install GPT4All desktop app & dwonload GPT4All Falcon wait until done to continue.

Dependencies

Install the GPT4All Library

python -m pip install gpt4all

Install the OpenAI Whisper Library

python -m pip install openai-whisper

Install the Speech Recognition Library

python -m pip install SpeechRecognition

Install the PlaySound Library

python -m pip install playsound

Install the PyAudio Library

python -m pip install PyAudio

Install the Sound File Library

python -m pip install soundfile

Install ffmpeg for Windows using Chocolatey

choco install ffmpeg

For MacOS Users

brew install portaudio

For Windows/Linux Users

python -m pip install pyttsx3

Configuration

To use Friday you need to app your OpenAI API Key openai.api_key = "KEY_HERE"

The duration variable can be raised for more time to respond to the AI. Default is 2 seconds

def start_listening():
    with source as s:
        r.adjust_for_ambient_noise(s, duration=2)
        print('\nSay', wake_word, 'to get started. \n')

Got an issue? Submit a pull request or submit an issue!

Developed by Toasty | Facekick x Dawnstar