摘要 |
PURPOSE:To easily impart accent information to synthesized voices without requiring special knowledge by imparting the accent information of registered natural voices to words or clauses at the time of synthesizing arbitrary characters. CONSTITUTION:The accent information of the registered natural voices is imparted to the words or clauses at the time of synthesizing the arbitrary characters. Namely, voice input is executed separately from character input and the accents of the voice input are used when the similarity of the spectral information of both inputs is large. Then, the accents to be added to the synthesized sounds by the voice input are set and the easy addition of the accents to the words not included in the conventional accent dictionary or the words not included in an accent dictionary for dialects is possible. The more natural voice synthesis is thus possible without requiring special knowledge. |