video2text 0.1.0

A command-line tool to transcribe audio from YouTube and other sources using Whisper.

transcriber

Transcribes audio and video from a URL to text using OpenAI Whisper (local) with Python.

Quick Setup

Run the provided setup script (macOS/Homebrew or manual ffmpeg for others):

./setup.sh

This will:

• Install uv (if needed)
• Create and activate a virtual environment
• Install all Python dependencies
• Install ffmpeg (macOS/Homebrew) or prompt for manual install

Manual Setup (Alternative)

1. Install uv:

   pip install uv
   

2. Create and activate a virtual environment:

   uv venv .venv
   source .venv/bin/activate
   

3. Install dependencies (from project root):

   uv pip install -e .
   

4. Install ffmpeg (required for audio extraction):
   • macOS (Homebrew):

     brew install ffmpeg
     

   • Other OS: Download from ffmpeg.org and ensure ffmpeg and ffprobe are in your PATH.

Usage Example

from transcriber.main import transcribe_url, combine_transcription

url = "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
result = transcribe_url(url)
print(result['text'])

• transcribe_url(url) downloads and transcribes the audio from the URL.
• transcribe_file(filename) transcribes an audiofile.

Notes

• This package uses local Whisper (no API key required).
• The first run will download the Whisper model (hundreds of MB).
• ffmpeg must be installed and available in your system PATH.