GENERATING SPEECH PATTERN TEMPLATES

摘要

GENERATING SPEECH PATTERN TEMPLATES An arrangement for generating a speech pattern includes storing a plurality of reference speech pattern templates, e.g., words. Each template comprises a sequence Or acoustic feature signals representative of a prescribed spoken word of a first speaker. For each reference word template, a set of signals representative of a predetermined subword unit, e.g., demisyllable in said reference word template is also stored. An utterance or a reference word by another speaker is analyzed to form a time frame sequence of acoustic feature signals for the reference word. A set of signals representative Or the time alignment between the stored reference word feature signals of the first speaker and utterance feature signals oi the other speaker is generated. The sequence of utterance feature signals of the other speaker corresponding to the predetermined subword unit is selected responsive to the stored reference word predetermined subword unit representative signal and the time alignment signals.