AI Everywhere YosoyX Looking At Tech

It has been a long-standing goal for the Division of Information Technology to allow Texas State University to provide an inexpensive method for captions and transcript generation for video and audio files used to support multiple activities across the institution. These activities include marketing, archive videos, and, most importantly, instructional media. Captions and transcripts help the university address ADA requirements, while also providing support for multiple learning styles.

The main barrier for a university-wide implementation of a captioning/transcription service has been cost. Under the current implementation, the captioning/transcription services available to the university community cost $60 per hour of video.  These services rely on human captioners.  This makes a system-wide adoption of captioning extremely costly.

The Technology Innovation Office embarked on a study to determine the cost of implementing an AI-driven platform that would leverage speech to text services to produce captions. The team put a random set of previously captioned files through the AI-driven speech to text services provided by AWS, Google Cloud Platform, IBM/Watson and Microsoft Azure, deriving accuracy and cost measures from the output. This assessment is considered as Phase 1 of an initiative that can lead to an implementation process, Phase 2, where an AI-driven captioning service would be integrated with MediaFlo to automatically create captions for all uploaded media. This service will also provide support different learning styles by providing transcripts of videos, which can be delivered to users via the video delivery platform.