We propose to develop a system for dissemination and retrieval of information about the usefulness of computerized voice and speech analysis and synthesis systems. Comparative studies of commercial and public domain analysis and synthesis packages will be performed. A seminar will be organized to provide a forum for discussion of computerized voice analysis and synthesis. The seminar will be designed to have broad appeal in the hope of attracting engineers, business people, and the general public, as well as professionals who have special demands and needs for speech and voice (singers, actors, public speakers, teachers, lawyers, etc.). A voice data and information service will be developed and maintained on- line for access by scientists over a computer network. The service will include a voice bulletin board, event calendar, agency information, equipment information, and speech and voice analysis and synthesis programs and data. Voice analysis would be greatly facilitated by the development of standards for data input and output. Some consensus already exists, since most of the major vendors support 16-bit integer data input files. However, file headers are generally different. Our analysis (GLIMPES, CSPEECH) and synthesis (SPEAK) packages will be modified to support a variety of file formats (ILS, Haskins, ASCII, integer binary, etc.). Standards for sampled data file input and analysis parameter output will be developed and disseminated.
Showing the most recent 10 out of 151 publications
