eduzhai > Applied Sciences > Engineering >

Speaker Independent and Multilingual/Mixlingual Speech-Driven Talking Head Generation Using Phonetic Posteriorgrams

  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Generating 3D speech-driven talking head has received more and more attentionin recent years. Recent approaches mainly have following limitations: 1) mostspeaker-independent methods need handcrafted features that are time-consumingto design or unreliable; 2) there is no convincing method to supportmultilingual or mixlingual speech as input. In this work, we propose a novelapproach using phonetic posteriorgrams (PPG). In this way, our method doesn tneed hand-crafted features and is more robust to noise compared to recentapproaches. Furthermore, our method can support multilingual speech as input bybuilding a universal phoneme space. As far as we know, our model is the firstto support multilingual mixlingual speech as input with convincing results.Objective and subjective experiments have shown that our model can generatehigh quality animations given speech from unseen languages or speakers and berobust to noise.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×