Complex waveform phonetic speech synthesis
Abstract
A new procedure for the synthesis of human speech by rule was presented. The technique has as its base the time domain specification of a complex waveform for each of the phonemes. Distinct waveforms are required for the prevocalic and postvocalic occurrences of some phonemes. The procedure is unique in that contextual variations are provided by a set of algorithms rather than simply concatenating the stored waveforms. Sonograms, time plots, and pitch contours of the natural and synthetic forms of several words are compared. Tests are conducted in which the intelligibility is approximately 90 percent with most listeners being impressed by the naturalness of the synthesized speech. Improvements are suggested to increase the intelligibility and naturalness of the synthesizer.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- 1974
- Bibcode:
- 1974PhDT........63H
- Keywords:
-
- Phonetics;
- Sound Waves;
- Speech Recognition;
- Waveforms;
- Acoustics;
- Algorithms;
- Auditory Perception;
- Intelligibility;
- Phonemes;
- Words (Language);
- Communications and Radar