Visit ChatPod at https://chatpod.onrender.com (Currently inactive, ping me if you would like to view the demo)
- With the increasing popularity of podcasts, we recognized the need for a more efficient way to search through and extract relevant information from these audio files.
- This project (ChatPod) leverages the power of ChatGPT to accurately process natural language queries and retrieve relevant portions of the podcast transcripts.
- Feedparser: RSS feed processing
- Whisper: Voice-to-text
- FAISS: Vectorstore
- LangChain: Overall backend framework
- OpenAI ChatGPT: Large language model (
gpt-turbo-3.5) - Streamlit: Frontend interface
- Render: Deployment of Dockerized app (cron-job.org to keep instance active)
- Create Upload page (pipeline from audio to chat over vectorstore)