APPARATUS AND METHOD FOR SELECTING THE UNITS IN A CORPUS-BASED SPEECH SYNTHESIS

摘要

1. This invention belongs to the art described in the claims ; Corpus-based synthesis unit of the present invention is selected from the voice synthesizer device and a method and a readable recording medium having thereon a computer program for realizing the method thing of. ; 2. Invention by selecting a technical problem ; The invention Corpus-based speech synthesis (TTS) synthesis columns with the lowest error distance between the synthetic unit in trying to solve this, between the synthesis speech synthesis unit appearing in reducing the error, so as to enable selective combining unit for overall natural speech synthesis apparatus and method and, to provide a recording medium readable by the computer the method in which a program for realizing that its purpose is. ; 3. The gist of the solution of the invention ; The present invention, in corpus-based speech synthesis unit selector at the synthesizer, using the most likely candidate units present in the speech database and the target unit distance of speech feature vectors but selected by the target cost calculation means for selecting the speech synthesis unit is continuously obtained from the same unit as the candidate; And receiving the candidate unit, the connecting unit between the intensity, pitch, but the connection costs calculated using the spectral distance, by varying the spectral distance calculation for each of the combining unit creates heat, that in the smallest cumulative distance synthesis unit heat The calculation includes the connection costs and the final decision means for selecting synthesis units. ; 4. An important use ; The present invention used to fight and corpus-based speech synthesizer of the invention.