This project is part of the course Building AI products with OpenAI taught by Sidharth Ramachandran.
In this project, I built an LLM app that summarizes a podcast episode, identifies podcast guests, and key momments. You can view it at https://podcast-summarizer.streamlit.app/.
-
Part 1: use a Large Language Model (LLM) from OpenAI to build the information extraction functionality paired with a Speech to Text model for transcribing the podcast.
- I used Whisper as the speech to text model.
- I used the OpenAI
gpt-3.5-turbo-16kmodel to generate the summary by passing in the generated transcript.
-
Part 2: use a simple cloud deployment provider to easily convert the information extraction function to run on demand - this would be the app backend. See Modal.
-
Part 3: use ChatGPT from OpenAI as coding assistant to create and deploy a front-end that allows users to experience the end to end functionality. See streamlit.io.