Automating Lecture Video Speech Recognition

SpokenMedia was an umbrella project for the development of tools and services to enable rich media notebooks for learning and teaching.

The key enabler for these rich media notebooks are full text transcripts of academic video lectures. Interactive text transcripts enable better searching and organizing of video lectures, and make possible their combination with text-based materials. In the case of MIT, imagine combining video lectures from courses with lecture notes or exams at MIT OpenCourseWare.

The SpokenMedia project demonstrated a couple tools and services:

• automatic lecture transcription (with potential integration in larger media production workflows),
• a video player with a video-linked transcript (and other interactive features), and
• a transcript editor.

The SpokenMedia Project was supported in part by the MIT / Microsoft iCampus Alliance. SpokenMedia took as inspiration two prior iCampus projects—it builds on the potential of automatically transcribing video lectures from the Spoken Lecture project and the use of video clips in education from the Cross Media Annotation System project.