DEMONSTRATION - EMOTIONAL SPEECH SYNTHESIS

Generally, in the speech synthesis system, it makes models by learning the "normal" speech data base.
However, if the system learns more emotional speech, what speech is synthesized?
This section demonstrates synthesized speech in the case of learning "angry" speech.

• Examples of speech used to learn
  • "Arayurugenjitsuwo, subetejibunnohouenejimagetanoda."
    ( Every fact was distorted all to me.)
    Normal Speech
  • "Oi! baachammadashindeneenoni, nantekotosundayo!"
    (Hey! That's a awful thing to do! My grandmother hasn't die yet!)
    Angry Speech

• Synthesized speech
  • "Ochitsuiteirareruwakenaidaro!"
    (How can I calm down myself?)
    Normal Speech Angry Speech
  • "Nandesonnakotoshinakyaikenaindayo!"
    (Why must I do that?)
    Normal Speech Angry Speech
  • "Issyuukanbakari, New York wosyuzaishita."
    (I stayed in New York for coverage for a week.)
    Normal Speech Angry Speech

As the last example, we can synthesize "angry" speech from even normal speech sentence by using this technique.
This technique allows us to expect synthesizing speech with more human emotion, and it is possible to apply this technique to the voice actor application and the automatic dubbing system and so on.