Speech and Language Technology for Disabled Persons
ASEL is looking into methods for creating a high quality, low cost speech synthesizer to be incorporated into communication devices for people with severe speech impairments. To this end, optimal coding methods for storing diphones are being investigated. The method discussed here isformant coding for use in a parallel formant synthesizer This method will allow for natural sounding, age and gender appropriate speech output with an unlimited vocabulary. In addition, since the data will be coded into frequency domain parameters, prosodic information can be easily manipulated to further improve the naturalness and intelligibility of the speech. The method will be evaluated on the basis of the quality of the synthesized output that it produces, the compactness of coded speech information, and the facility with which prosodic features can be manipulated (e.g. pitch, duration and amplitude). This paper will focus on the method for obtaining control parameters for the parallel synthesizer.
Bibliographic reference. Samworth, Katherine T. (1993): "A method for obtaining control parameters for a parallel formant synthesizer", In SLTDP-93, 161-164.