High-Speed Speech Calls for Hardware
From: Pittsburgh Post-Gazette - 09/20/2006
By: David Templeton

Carnegie Mellon University electrical and computer engineering professor Rob
Rutenbar seeks to improve the capabilities of speech recognition technology
by inventing better hardware, namely a computer chip that can comprehend
speech and process it faster than current software. This is important,
because real-time speech recognition software is about to hit a wall. The
work of Rutenbar's CMU research team has thus far yielded a prototype chip
that can recognize 1,000 words, albeit not in real-time speed. Rutenbar's
long-term objective is a chip capable of understanding 50,000 words at a rate
that beats real time 1,000-fold, a breakthrough that would aid national
security initiatives such as wiretap analysis. The technology's potential
commercial applications include improved cell phone voice recognition.
Stanford University electrical engineering professor Teresa Meng lauds
Rutenbar's effort as the most sophisticated speech recognition research
endeavor around, noting that Rutenbar has "put grammar and structure in the
chip in a multistep recognition process to cast a fairly wealthy set of
thinking into hardware." Rutenbar is developing technology to assist Homeland
Security through grants from the Defense Department, the semiconductor
industry, and the National Science Foundation. 

Read the entire article at:
http://www.post-gazette.com/pg/06263/723150-96.stm

Links:
Rob A. Rutenbar
http://www.ece.cmu.edu/directory/details/150/
http://www.ece.cmu.edu/~rutenbar/
http://c2s2.ece.cmu.edu/team/bio/rutenbar.php

Carnegie Mellon Engineering Researchers To Create Speech Recognition in Silicon
http://www.cmu.edu/PR/releases04/040913_speech.html

Chips promise to boost speech recognition
http://news.zdnet.com/2100-9584_22-6108417.html

Teresa Meng
http://dualist.stanford.edu/~thm/