VerBall

Inspiration

Manually recording game statistics with pen and paper is inefficient and inconvenient. Fans who cannot attend games in person due to personal circumstance or accessibility barriers may miss out on the exciting fan experience of following a live, play-by-play game.

What it does

Takes spoken sports commentary and turns it into accessible visualizations and live, play-by-play updates in our VerBall app for teams to use for stat tracking and fans to follow along anywhere, anytime.

How we built it

Speech-to-text via Deepgram's Nova-3 API Event extraction using Anthropic's Claude Haiku 5.5 LLM to identify players and game events to turn into live game statistics

Challenges we ran into

Coming up with a strong idea that extended beyond a simple stat tracking app that could benefit players, coaches, and fans Speech-to-text transcription would commonly transcribe player names and basketball terms incorrectly (e.g. "layup" becomes "lab") which made then hard to track into stats

Accomplishments that we're proud of

Combining game tracking, statistical analysis, and live updates into one consolidated platform Successfully extracting multiple trackable events form longer segments of spoken commentary Designing a shared experience that helps teams analyze performance while allowing fans to follow games remotely

What we learned

Speech-to-text and event extraction using AI Designing for different audiences meant learning to balance accurate data collection with clear, accessible visualizations

What's next for VerBall

Expansion beyond basketball into other sports that use spoken commentary Monetization potential by offering subscription plans for teams to have saved game history and advanced analytics Improving accuracy with better player name recognition, sport-specific terminology, and multiple events within fast-paced commentary