talk2pdf 0.3.0

No project description provided

talk2pdf

talk2pdf converts videos of talks into a PDF with the transcript and associated screenshots.

Video	Result

Prerequisites

talk2pdf uses OpenAI's APIs -- you will need an OpenAI API key.

Note Your API key is ONLY used to make OpenAPI calls.

1. Create an OpenAI account
2. Go to platform.openai.com
   1. Click on your account in the top right
   2. Click View API Keys
   3. Click "Create new secret key"
   4. You will need to provide this API key to talk2pdf (see below)

talk2pdf uses FFmpeg for video/audio operations and yt-dlp to download video from online streaming services.

brew install ffmpeg yt-dlp

apt-get install ffmpeg yt-dlp

Running

Warning This will cost ~$0.50 / hr of video

This will generate a PDF in the current directory:

With a youtube video

pipenv shell
pip install talk2pdf

export OPENAPI_SECRET=sk-...
python -m talk2pdf <youtube-url>

Or, with an existing video file

pipenv shell
pip install talk2pdf

export OPENAPI_SECRET=sk-...
python -m talk2pdf <video-file>

Configuration

talk2pdf looks in TALK2PDF_CONFIG_DIR, then XDG_CONFIG_HOME/talk2pdf, and finally $HOME/.config/talk2pdf for a config.json file. You can put your OpenAI key in there instead of setting it in the environment

{
  "openapi_secret": "sk-..."
}

talk2pdf uses TALK2PDF_CACHE_DIR, or XDG_CACHE_HOME/talk2pdf, or finally $HOME/.cache/talk2pdf as a cache directory. This is where intermediate files are stored.

Contributing

pipenv shell
pip install --editable .
python -m talk2pdf ...

pip install --upgrade build twine
rm -rf dist
python -m build
python3 -m twine upload dist/*

Roadmap

• add timestamps to each paragraph