From Carolyn Kotlas at TL INFOBITS:
This fall researchers at the Massachusetts Institute of Technology released the MIT Lecture Browser, “a web interface to video recordings of lectures and seminars that have been indexed using automatic speech recognition technology.” Users can search on terms or phrases and then play the video at the point(s) in the recording where their search term appears.
This technology first involved creating software that converted audio to text. Next the software was trained “to understand particular accents using accurate transcriptions of short snippets of recorded speech.” Then the researchers provided data on uncommon words so the software could recognize technical terms that might be used in university lectures.
While the transcript’s accuracy can be affected by speakers’ verbal pauses or by nonnative English speakers’ accents, the texts can be very close to the audio originals. The transcripts’ accuracy is sufficient for searches, and there is potential for use by hearing-impaired students if future plans to allow users to make corrections to transcripts are implemented.
You can search and try out the Lecture Browser at http://web.sls.csail.mit.edu/lectures/.
I did the obvious search for “library” and got hits in 12 lectures; they weren’t the faculty lectures I was expecting, but talks gave by visiting folks, like Maurice Sendak and Bill Gates. The results show you the transcript of the talk with KWIC, and allows you to easily jump to the relevant section of the talk. Couldn’t view any of the video on FireFox, but it worked well in IE. Here’s a shot of Maurice Sendak halfway through his talk (click to embiggen and see the text)
Technorati Tags: MIT