[2O1-GS-7] Vision, speech media processing: generation

[2O1-GS-7-04] Emotionally Conditioned TTS with Facial Expression

〇Jun Enoki¹, Takayuki Suzuki¹ (1. Opus Inc.)

Keywords:text to speech, Arts and entertainment applications

In these days, generated speech by the modern TTS system can be undistinguishable with real human's speech, and many researches have been studied even on emotionally conditioned TTS.
Here we explore another way to control emotions on speech synthesis by combining facial expression data to achieve intuitive conditioning.
In this paper, we share our experimental results and discuss the details.