Speaker Identification in Television Closed-Captions Docket: UAH-P-13010 Technology Television channels generate a tremendous amount of useful information each day, especially news channels. The busy nature of today's society coupled with consumer preference, both on television and the Internet, result in many of the stories being missed or overlooked. Currently there is no easy way to browse for these on the Internet. Researchers at UAH have developed a method to generate full transcripts, summaries, and key-word tags of news stories using the closed captioning. The technology uses machine learning techniques to identify the speaker—a function not performed in closed captioning—and captures dialogue from closed captioning. Once the information is combined, a transcript is created, and a summary is made available on a website. The generated summaries will be posted to a website allowing the user to efficiently browse their favorite news topics while having access to other stories they may otherwise miss. Over the course of using the website, the user's interests will be detected and suggestions will be made leading them to other similar articles. Applications Phone apps for news and entertainment websites News websites Entertainment websites Political organizations Archiving and data mining services DVR functions Advantages Provides video and closed-caption news summaries at minimal cost Simplifies data mining and search functions Ability to detect and suggest stories and links Commentators' opinions can be categorized and browsed Links to videos Status State of Development: Prototype Licensing Status: Available for licensing Patent Status: Patented