| 摘要 |
The invention makes it possible to synthesize any acoustic data by concatenation of individual audiosegment zones, the instant at which the concatenation of two successive audiosegment zones is carried out being chosen in accordance with properties of said audiosegments. In this manner synthesized acoustic data can be generated which after conversion into acoustic signals do not differ from the corresponding naturally produced acoustic signals. The invention notably makes it possible for synthesized speech data to be generated taking into account coarticulatory effects, by concatenation of individual speech-audiosegments. The speech data provided in this way can be converted into speech signals which are indistinguishable from natural spoken speech.
|
