Media Lab Speechhome Project
The Media Lab’s Human Speechome Project, led by Associate Professor Deb Roy, has been amassing several terabytes per week of digital audio and video (A/V) recordings of early childhood learning and socialization data. These massive quantities of A/V data will be processed and analyzed using a suite of innovative data mining tools that Professor Roy and his team have been developing. By mid-2008, the information will have been assembled into a database exceeding one petabyte (1,000 terabytes) in total capacity. The speech and video data will be processed and analyzed by several hundred parallel processing devices in one of the most extensive scientific analyses of long-term infant learning patterns ever undertaken. Speech and video mining technologies emerging from this research will impact multimedia data management, business intelligence, and securities industries.
When fully built out, the Human Speechome Project computing infrastructure is expected to be composed of more than 3,000 Seagate SATA drives, more than 300 Hammer Z- Rack storage enclosures, more than 100 Marvell-based 10G/GbE switches, and about 400 blade processors. High-performance storage I/O anticipates the processing of 700 terabytes of data during each 12-hour overnight analytical run. To achieve the desired performance requirements, 150-drive stripes (aggregated virtual volumes) will be created using the native virtualization capabilities of Z-SAN. Protection against data loss will be delivered through RAID 10 mirrors (duplicate copies) of the raw video data, transform data, and metadata files.
MIT , http://www.media.mit.edu/press/speechome/speechome-sponsor.pdf
MIT Press Archive, Speechome, http://www.media.mit.edu/press/speechome/
MIT Media Lab project explores language acquisition, Press Release, May 12, 2006. http://www.media.mit.edu/press/speechome/speechome-mit.pdf
Celeste Biever, Watch language grow in the 'Baby Brother' house, New Scientist, 14:18 15 May 2006. http://www.newscientisttech.com/article/dn9167-watch-language-grow-in-the-baby-brother-house.html
The e-Science 2006 conference is designed to bring together developers and users of e-Science applications and enabling IT technologies from leading international and interdisciplinary research communities. The conference serves as a forum to present the results of the latest research and product/tool developments, and highlight related activities from around the world.