Revolutionizing sports highlight creation with the power of AI! Whether you're a broadcaster or content creator, our solution delivers engaging Olympic moments from a simple prompt. This project has won First Prize in Cerebral Beach Hacks in Sports Category.

https://devpost.com/software/olympic-highlight-generator

https://youtu.be/hxt9041EGsY?si=Bx5IOA2FI7Qg7J-4

The 2024 Summer Olympics is a global event, watched by millions worldwide. Traditional highlight generation is time-consuming and requires manual editing. We sought to create a tool that:

• Enables fans, broadcasters, and content creators to relive thrilling Olympic moments.
• Streamlines the production of personalized and professional-quality sports highlights.
  Our goal is to empower producers with AI to create engaging commentary and visuals tailored to consumer interests.

The Olympic Highlight Generator creates dynamic highlight reels from user prompts.
Key features:

1. Takes a natural language prompt as input.
2. Extracts relevant Olympic footage using AI-powered video search.
3. Generates an engaging script contextualized to the selected clips.
4. Converts the script into a realistic voice-over.
5. Automatically edits visuals to synchronize with the generated narration.

The result? 90-second professional-quality highlight reels delivered seamlessly.

Our project integrates multiple AI models and APIs in a unified workflow:

1. Video Search: User prompts guide the Marengo model (TwelveLabs) to extract relevant Olympic clips from a curated database.
2. Script Generation: The Pegasus model generates commentary tailored to the video context.
3. Text-to-Speech: ElevenLabs' API creates lifelike voice-overs from the script.
4. Video Editing: A custom algorithm aligns visuals with narration, ensuring perfect synchronization.
5. Web Interface: A user-friendly platform allows seamless interaction with the system.

• Ensuring generated scripts align with available Olympic footage.
• Synchronizing visuals with AI-generated voice-overs required meticulous model fine-tuning.
• Handling the integration of multiple API calls into a seamless pipeline.

• Fully automated a traditionally manual, labor-intensive process of highlight generation.
• Achieved high-quality voice-overs synchronized perfectly with footage.
• Created a scalable, user-friendly web platform.

• Advanced video editing and synchronization with AI-generated scripts.
• Integrating diverse APIs into a cohesive pipeline.
• Tackling unique challenges associated with working with Olympic footage.

• Live Event Highlights: Support real-time highlight generation during live Olympic broadcasts.
• Enhanced Personalization: Adjust tone, emphasize specific athletes or events, and more.
• Expand Video Index: Include a broader range of sports and other global events.

• Frontend: React.js (for user interaction)
• Backend: Node.js, Express.js (for API calls and integrations)
• AI Models:
  • Marengo (TwelveLabs): Video search and clip extraction.
  • Pegasus: Script generation.
  • ElevenLabs: Text-to-speech voice-over.
• Video Editing: Custom Python algorithm for synchronization.

• Darshan Rao
• Ayan Bhowmick
• Anupam Patil
• Shardul Datar
• Shaunak Mahajan