Publication Date
Spring 2024
Degree Type
Master's Project
Degree Name
Master of Science in Computer Science (MSCS)
Department
Computer Science
First Advisor
Robert Chun
Second Advisor
Thomas Austin
Third Advisor
Aditya Chhikara
Keywords
generative artificial intelligence, sports commentary, formula 1, modern transformer, T5 model
Abstract
This project aims to create a high-quality commentary generation system utilizing cutting-edge Generative AI technologies, with a particular focus on the T5 transformer-based text-to-text transfer transformer (T5). The primary goal is to create a fully autonomous and contextually aware commentary system that will be able to provide consistent and insightful commentary on dynamic events that reflect the level of detail normally associated with human commentary. Upon giving the input as a text input of the race events the model using large language models that are trained on a large range of datasets creates text-based commentary. The system in order to create human-level commentary in real-time modifies itself based on complex patterns along with contextual information. This approach can also be used in other real-time applications such as any number of indoor and outdoor sports cultural performances, policy debates, emergency broadcasting, etc. By providing individualized and multilingual commentary, the creative use of generative models has the potential to revolutionize sports broadcasting technology and improve the viewing experience.
Recommended Citation
Moliya, Hiral, "Formula 1 Commentary Generator using Generative Artificial Intelligence" (2024). Master's Projects. 1367.
https://scholarworks.sjsu.edu/etd_projects/1367
