AI Sports Recap

AI Sports Recap is a Streamlit-based application designed to generate video highlights and textual summaries of sports press conferences given a YouTube video link. The app leverages cutting-edge technologies including OpenAI's GPT-4o, Pegasus1 by TwelveLabs, and Docker for seamless integration and deployment.

Tech Stack

• OpenAI GPT-4o: Used for generating textual summaries and extracting relevant video clips based on the input query.
• Pegasus1 by TwelveLabs: Utilized for transcript generation and video clipping.
• Streamlit: Provides the UI and frontend for user interaction.
• Docker: Ensures a consistent and reproducible environment for running the application.

Features

1. Transcript Generation: Automatically generates a transcript of the provided video.
2. Highlight Extraction: Identifies and extracts key video segments that are relevant to the user's query.
3. Summary Generation: Creates a concise summary of the press conference.
4. Social Media Sharing: Allows users to share the generated summary and highlights on social media platforms.

Installation

Prerequisites

• Docker
• OpenAI API Key
• TwelveLabs API Key

Steps

1. Clone the repository

   git clone https://github.com/prateekchhikara/sports-highlights.git

2. Set up environment variables Create a .env file in the root directory and add your API keys:

   OPENAI_API_KEY=your_openai_api_key
   INDEX_ID=your_index_id
   TLABS=twelve-labs_api_key

3. Build and run the Docker container

   docker compose up --build
   

4. Access the application Open your browser and go to http://localhost:8501.

Usage

1. Enter the URL: Paste the YouTube video link of the press conference.
2. Select an option: Choose from predefined video options if available.
3. Submit a query: Input your question or query related to the video content.
4. View the results: The app will display the video highlights and textual summary based on your query.
5. Share on social media: Use the provided links to share the summary and highlights on Facebook and Twitter.

Code Overview

The main components of the application are as follows:

• Streamlit Frontend: Handles user input and displays results.
• Backend Functions:
  • generate_transcript: Uses Pegasus1 to generate video transcripts.
  • get_intervals: Extracts relevant video segments based on the GPT-4o output.
  • get_text_from_gpt: Queries GPT-4o to extract relevant content & their timestamps from the generated transcripts based on the user's prompt.
  • get_clippings_from_intervals: Generates relevant clips of the original video based on the identified time intervals & merges them into a single highlight video.
  • get_summary_and_title_from_gpt: Generates a summary and title for the video content.
• Social Media Integration: Provides links for sharing the content on social media platforms.

Example

1. User Interface:

   • Enter the URL of the video or select from predefined options.
   • Submit a question or query about the press conference.

2. Backend Processing:

   • Generate transcript of the video.
   • Use GPT-4o to summarize the transcript and identify key video segments.
   • Clip the relevant video segments and generate a consolidated highlight video.
   • Display the textual summary and video highlights to the user.

3. Sharing:

   • Users can share the generated content on Facebook and Twitter directly from the app.

Acknowledgements

• OpenAI for their powerful GPT-4o model.
• TwelveLabs for Pegasus1 API.
• Streamlit for the user-friendly frontend framework.
• Docker for the containerization and ease of deployment.

Contributors

Our project wouldn't be possible without the contributions of these amazing people! Thank you all for making this project better.