System for concatenation of audio segments in correct co-articulation for generating synthesized acoustic data with train of phoneme units

摘要

In the generation of synthesized acoustic data at least two audio segments, reproducing phoneme units, are selected from a data bank of acoustic data to be synthesized. Each audio segment has at least one single articulation region. The moment of concatenation, of a start of a used part of a second audio segment to the end of used part of first audio segment, is set in a region beginning directly before the used part of the second audio segment and ends after the first single articulation section of the used part of the second audio segment. This is in dependence on the properties of the used part of the second audio segment. Independent claims are included for a generator of synthesized acoustic data and for synthesized speech signals, and data memory.