A website to display book recommendations from the Lex Fridman Podcast.
Final project for CS 1699 Practical AI, Spring 2023.
cd src/
pip install -r requirements.txt
python -m spacy download en_core_web_sm
python -m spacy download en_core_web_lg
cp src/example.env
cp src/.env
Fill in required secrets in .env
file
./download_transcripts.sh
# Conver data to text format (removing timing informatio from *.vtt file)
./convert_all.sh
cd src/
./main.py
npm install --prefix site/
npm start --prefix site/
docker run --name site --rm -it $(docker build -q .)
- OpenAI Whisper integration
- Add logic to be alerted of a new podcast post (likely from RSS feed)
- Host on Google Cloud
- Create Dockerfile
- Automatic triggers and builds on pushes to main
- Run container on GC
- Run cron job to check for new podcast
- Remove duplicate posts
- Increase model accuracy
- Look into a case-insensitve model that does not rely on capitalization (this is bottlenecked by Whisper)
- "Capitalization normalization" did not work
- Categorizing recommendations
- Add genre information to each book
- Create running lists of reccomended books. This will include a "reccomended_in" with each podcast it was mentioned
- Setup
env
for API keys