GPTranscribe

*MacOs and Web App in progress
Welcome to GPTranscribe(for now)! This python application(for now) allows for transcription of system audio(ie a meeting or youtube video), extraction of important terms and GPT driven definitions for the terms you used to google during a meeting or video. Think of this as the equivelant of left clicking text and searching for a definition but for system audio!

Setup

GPTranscribe requires 3 main dependencies

1. python 3.7-3.9
2. An openAI API key
3. blackhole 2 Channel audio driver
4. ffmpeg

Assuming the user has 1, 2 and a cloned repo follow these steps for setup:

1. Install audio driver dependencies
   • blackhole2ch brew install blackhole2-ch
   • PortAudio brew install portaudio
2. Create a multi-audio device for blackhole + your current/desired output. Check here for instructions.
3. Create a .env file with OPENAI_KEY=your_api_key
4. Install python requirements pip install -r requirements.txt
5. Ensure you're using the multi-audio device from (2) and run python smart_transcribe.py
6. *optional: For a basic user interface run python server.py alongside