Ultra-wideband sensing of speech production

Ultra-wideband (UWB) technology has been identified as an attractive solution for certain biomedical sensing applications due to its safety, no requirement of physical contact, real-time capability, and low cost. This work considers UWB sensing of human speech production, potentially allowing significant improvements in synthetic speech production, speech pathology, and speech recognition. Results of a simple speech recognition experiment are presented. A multi-section cylindrical waveguide model is proposed to approximate the human vocal tract for future inverse scatting studies.