Speech Synthesis Using Sum of Frequency Modulated Sinusoids.

The purpose of this study was to investigate speech synthesis using a sum of up to 16 frequency modulated sinusoids. Two forms of utterances were synthesized sustained phonetic phonemes and complete words. The words were constructed by splicing together the sustained phonetic phoneme utterances of proper lengths. While toll quality speech was not produced, synthesized utterances were of high speech quality and were readily intelligible. This approach to speech synthesis offers several opportunities for future research such as using additional oscillators, low bit rate speech transmission, and possibly toll quality speech. Keywords include Speech Synthesis, Synthesis, Sum of sinusoids, and Spectral speech synthesis.