Multi-Sensor Noise Suppression and Bandwidth Extension for Enhancement of Speech

To my dear family, friends and my teachers iii ACKNOWLEDGEMENTS First and foremost, I would like to express my deepest gratitude to my advisor, Dr. David Anderson, for the guidance, support, endurance, and encouragement during my graduate studies at Georgia Tech. He has always been supportive of me not only when I made progress but also when I made mistakes. I am also deeply impressed with his dedication to research and students. What I have learned from him are by no means limited to signal processing techniques, but rather applicable to all the life. His positive influence will no doubt propagate beyond one graduate degree and serve me well for a long time. I also would like to thank Dr. Bhiksha Raj, who was my host at Mitsubishi Electric Research Lab. My knowledge was enhanced and my horizon was expanded after each technical conversation with him. I have also been greatly benefited from the interactions with Dr. Mark Clements, and I am thankful to his guidance. More than four years of graduate studies at Georgia Tech provided me with the wonderful opportunity to work closely with several colleagues and co-researchers. I cherish the fruitful interactions I had with the researchers that I worked with on the DARPA supported low bit-rate speech coding research project. I also thank all my colleagues in the Efficient Signal Processing (ESP) Lab and Cooperative Analog and Digital Signal Processing (CADSP) group for their collaboration in several research adventures. I am also thankful to my friends outside Georgia Tech for their encouragement during the past four years. My family members mean a lot to me. Especially, I am grateful to my eldest brother for his moral support. I also thank the fellows at Mitsubishi Electric Research Lab.