Podcast QA CLI Tool

The Podcast QA command line tool offers an innovative way to explore the knowledge embedded in some of the world's most influential podcasts. By simply asking a question, you can get actionable insights and references to relevant resources. The sources of these insights are the analyzed content from the supported podcasts.

Currently, our supported podcasts include:

HubermanLab podcast - hosted by Andrew Huberman, a neuroscientist and tenured Professor at the Department of Neurobiology at Stanford University School of Medicine.

Features

The CLI tool allows you to:

Setup output paths: Define the directories where the outputs will be saved. If the paths are not set up, the tool will use default paths.
Perform Question Answering: In this mode, the tool will provide an answer to your query and point out relevant segments from the podcast episodes.
Search for resources: If you're interested in gathering resources without checking their relevance or obtaining a specific answer, this mode will fetch and return the relevant podcast episodes and segments.

Note: In the resource searching mode, the tool will not require calling the OpenAI API, thus it won't incur any costs and won't require providing the OPENAI_API_KEY.

Installation

There are two ways to install the CLI tool: directly from PyPi using pip, or by cloning the repository and installing the requirements.

Method 1: PyPi

pip install podcastqa

Method 2: From Repository

First, clone the repository to your local machine:

git clone https://github.com/DataScienceDisciple/podcast-qa
cd podcast-qa

Then install the necessary dependencies:

pip install -r requirements.txt

Environment Variables

Before you begin, you should set up your environment variables.

Create a .env file in your project root directory.
You can set environment variables through the CLI tool using the --set-env command. This command takes two arguments: the variable name and the variable value. For example, to set the OPENAI_KEY, run the following command:

python main.py --set-env OPENAI_API_KEY your_openai_key

Note: Replace your_openai_key with your actual OpenAI API key.

Usage

Setting up output directories

You can setup the directories to which the outputs will be saved. If the paths are not setup, the default paths will be taken.

python main.py --set-output-paths path/to/qa_output path/to/images

Question Answering Mode

In this mode, the CLI tool returns the answer and relevant segments for a given question.

python main.py --question "What is the impact of sleep on cognitive function?"

You can also customize the parameters of the QA engine:

python main.py --question "What is the impact of sleep on cognitive function?" --embedding_model sbert --n_search 20 --n_relevant_segments 3 --llm_model gpt-3.5-turbo --temperature 0

Resource Searching Mode

In this mode, the CLI tool returns only the relevant resources, without checking if they are relevant nor providing a final answer.

python main.py --resources "What is the impact of sleep on cognitive function?"

Notes

In the resource searching mode, the tool will not require calling the OpenAI API, thus it won't incur any costs and won't require providing the OPENAI_API_KEY.
The question answering mode requires calls to the OpenAI API to check the relevance of the segments and construct a final answer.