Detection of Nonstationary Noise and Improved Voice Activity Detection in an Automotive Hands - free Environment

Speech processing in the automotive environment is a challenging problem due to the presence of powerful and unpredictable nonstationary noise. This thesis addresses two detection problems involving both nonstationary noise signals and nonstationary desired signals. Two detectors are developed: one to detect passing vehicle noise in the presence of speech and one to detect speech in the presence of passing vehicle noise. The latter is then measured against a state-of-the-art voice activity detector used in telephony. The process of compiling a library of recordings in the automobile to facilitate this research is also detailed. ACKNOWLEDGEMENTS I owe a great deal of thanks to very many people for making my graduate work possible and enjoyable. I’d like to start by thanking my parents for their unwavering support and generous encouragement throughout my academic career and entire life. I’m certain I could not have gotten this far without it. Next I’d like to thank Rick Brown for his commitment, encouragement, assistance and company over an intense three years of electrical and computer engineering. I owe two of those three years to him convincing me that I should go to graduate school and, despite a few rough spots, I’m very glad he did. I don’t think it’s possible for me to quantify everything I’ve learned from him over that time or my appreciation for it. Bose Corporation deserves my thanks for their generosity in supporting me for my first year of graduate school. I’d specifically like to thank Jeff Faneuff for making my relationship with Bose possible. He and Vasu Iyengar also contributed much valuable advice and feedback on this work. I owe Jeremy Slater and David Feinzeig thanks for their support, their company, and their assistance with this project. They’ve made many late nights in the lab rather enjoyable and were anything but hesitant to aid in the data collection effort. Last, but certainly not least, I’d like to thank my committee, Brian King and John McNeill, for their time and their valuable feedback.