This is the first year of funding of a 4 year continuing award. This project will investigate new approaches to the classification, analysis, and resynthesis of sound. The research plan is composed of three elements: The majority of the effort will focus on the development of algorithms and techniques for sound parameterization and synthesis. The PI will explore ways to sort and match sounds with algorithms, voice algorithm parameters, and test human subjects to determine success of synthesis. He will develop classification and analysis software to automate existing algorithms, eliminating the human guidance andjudgment currently required. He will develop new algorithms for classes of sounds that do not yield to existing parametric synthesis algorithms. He will also work on novel interfaces for controlling sound, and create systems that can take advantage of flexible sound synthesis and control. By the end of the project, machine algorithms will have been developed that can take certain (large) classes of sounds as input, select the appropriate synthesis algorithm, extract parameters for resynthesis, and predict human responses as to the quality of the synthesized sound; simple systems will have been constructed which demonstrate the utility of the synthesis algorithms. This research will yield new tools at both the input and output ends of the interactive sound chain to benefit diverse applications such as portable computers, virtual reality, augmented reality, awareness systems, teleconferencing and remote collaboration applications.
