Speech synthesis with various voice characteristics and speaking styles

Speech synthesis with various speaking styles

Model Style dependent Style mixed
Reading
Rough
Joyful
Sad

Speaker adaptation for style mixed model

Speaker adaptation 1
Initial model Adapted model

Sentence: Sono fujin wa me wo kocchi ni mukete wa iru ga mite wa inai.
Reading Joyful Sad

Speaker adaptation 2
Initial model Adapted model

Sentence: Gakusei wa report wo oku to chotto atama wo sagete dete itta.
Rough Sad