This research project will
provide a quantum leap in access to multilingual cultural audiovisual archives
by integrating new methods in speech and language technology. It will result in
new approaches for use by catalogers and researchers that will substantially
reduce the cost of obtaining transcripts and metadata and will significantly
improve multilingual search of large audiovisual collections.
The project will deliver
new technologies in the areas of robust speech recognition and automated
cataloging in multiple languages. The VHF Oral History Collection contains vast
amounts of unconstrained, natural speech. It could well be the largest and most
complex single topic digital video library in the world. Providing access to this collection will put
the component technologies (especially automatic speech recognition,
concept-based indexing, text summarization, etc.) to the ultimate test. Beyond
the great importance of improving access to the VHF Oral History Collection
itself, we see great opportunities for further applications of our work.
We view this project as a
rare opportunity for research in human language technology. We have a vast
collection of multilingual, topical material from underserved portions of the
world’s population (including elders and speakers of minority languages); the
Shoah Foundation has an ongoing, labor-intensive manual indexing and annotation
effort into which we can directly incorporate our research results; we have
identified the technological improvements that are needed and we have assembled
a team with linguistic and technical expertise to realize them; and, finally,
we have a clearly identified community of users (historians and educational
programs) who will benefit immediately from our work. We expect that this novel
interweaving of human language technologies will serve as model for future
efforts in business and education. The technological innovations in MALACH,
with their focus on access to services, information, and education, will drive
further strategic research and development.