Diagnosing, projecting and training criterial function of speech synthesis by selecting units and apparatus for making the same

摘要

The invented method of diagnosing, projecting and training criterial function of speech synthesis by selecting units comprises a preparatory phase and a diagnostic phase. The a preparatory phase comprises in a first step annotation of synthetic speeches generated by a speech synthesis tested system based on a method of selection of units, wherein in the course of the annotation, there are identified faulty sections on the speeches. In the second step, it comprises a detection of esoteric and/or extreme values of the unit prices creating annotated synthetic speeches generated by the speech synthesis tested system and/or in the second step, it comprises a detection of objectively existing events in the synthetic speeches. In the course of the diagnostic phase, there is enumerated dependence between the positions of the esoteric and/or extreme values of the unit prices creating annotated synthetic speeches and sections in speeches identified in the course of annotation. Apparatus for making the above-described method is provided with control elements, a display output device, and a loudspeaker for post-synchronization of the text. It is further provided with a speech synthesis system, or let us say, a system for conversion of the text to speech comprising procedure for making the method.