Upon Phoneme Synthesis Based on Dynamic Semantic Modeling

摘要

The paper discusses the possibility of phoneme synthesis starting from a dynamic model of the semantic content. The meaning of words is modeled by a dynamic process of linear superposition of. the. component phonemes. The experimental results support the premise that the phonemes dynamics at the semantic level is conserved throughout the speech formation. Speech phonemes analysis suggests a dynamics typical for a controlled series of chaotic attractors. This is consistent with the proposed semantic model where the superposition of chaotic attractors determines a unique dynamic behavior of the resultant word. The proposed approach integrates the separate units of language coming in sequence into a unitary whole and can serve as basis for a new dynamic theory of language generation.