Turbo Processing of Speech Recognition
Utah State University researchers have discovered a way to combat speech recognition issues with the Turbo Processing System and are seeking an industry partner interested in licensing the technology. The turbo speech processing technology has demonstrated a substantial improvement compared to a speech recognition system that does not use it. Specifically, in a laboratory test the technology is able to reduce the relative error rate by up to 60%. The technology may be worked into existing speech recognition methods, including those that employ hidden Markov models, or neural nets.
Benefits
Advantages include:
- Low error rate
- Works in noisy environments
- Multi-stage processing for accuracy
- Compatible with legacy technology platforms
Applications
Potential applications include electronics, automotive, mobile phones, and medical equipment.
Contact
Questions about this technology including licensing availability can be directed to:
Alan Edwards, MA, JD
Manager, Technology Transfer Services
(435) 797-2328 alan.edwards@usu.edu
USU ID C11011
Inventors
Todd Moon, Ph.D., Electrical and Computer Engineering
Jacob Gunther, Ph.D., Electrical and Computer Engineering
Development Stage
TRL 4