Corise Podcast Frontend App

Overview

Corise Podcast Frontend App is a podcast feed parser application built using Streamlit, Python 3, Modal, and LLMs like GPT and Whisper. It is designed to parse RSS and summarize podcast episodes from https://www.listennotes.com/.

Features

• Parses RSS feeds and summarizes podcast episodes
• Retrieves podcast guest information using OpenAI and Google's Programmable Search Engine API

Technologies Used

• Streamlit
• Python 3
• Modal
• LLMs (GPT and Whisper)
• OpenAI
• Google Programmable Search Engine API

Installation & Deployment

Backend

1. Clone the repository
2. Set up a new account on Modal
3. Follow the set up steps in Modal Home to install the modal-client and create a set of tokens
4. Create 3 secrets on Modal to run the Podcast Backend:

Secret Name	Environment Variable Name	Description
my-openai-secret	OPENAI_API_KEY	An OpenAI API key with access to gpt-3.5-turbo-16k from OpenAI Platform
google-custom-search-api-key	GOOGLE_CUSTOM_SEARCH_API_KEY	A programmable Google Search API key. Follow the instructions at https://developers.google.com/custom-search/v1/introduction to set it up.
google-custom-search-engine-id	GOOGLE_CSE_ID	This must have a value assigned to the Search Engine ID project. Follow the instructions at https://developers.google.com/custom-search/v1/introduction to set it up.

5. Deploy the backend to Modal using modal deploy ./podcast_backend.py

Frontend

1. Go to https://share.streamlit.io/ and set up a new account
2. Login, then go to the Share Apps page to deploy a new application. Provide the following options:
   • Repository: https://github.com/KalleV/corise-podcast-frontend-app
   • Branch: main
   • Main file path: "podcast_frontend.py"

Usage

1. Enter an RSS feed URL in the input box
   • Examples:
     • https://feed.podbean.com/thecurioustanguero/feed.xml
     • https://maestrosdelamusica.libsyn.com/rss
2. Click on the "Summarize" button to generate a summary of the podcast episode
3. The app will also display information about the podcast guests

Modal Remote Function calls

Deploy backend to Modal

modal deploy ./podcast_backend.py

Running commands on Modal

Note: the /content/podcast/ is a path inside the image rather than local

modal run ./podcast_backend.py --url https://access.acast.com/rss/d556eb54-6160-4c85-95f4-47d9f5216c49 --path /content/podcast/

Download a transcription using a remote Modal function

python3 transcribe_podcast.py

Fetch podcast guest information using remote Modal function

python3 get_podcast_guest.py

License

This project is licensed under the MIT License - see the LICENSE file for details.