eduzhai > Applied Sciences > Engineering >

Peking Opera Synthesis via Duration Informed Attention Network

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Peking Opera has been the most dominant form of Chinese performing art sincearound 200 years ago. A Peking Opera singer usually exhibits a very strongpersonal style via introducing improvisation and expressiveness on stage whichleads the actual rhythm and pitch contour to deviate significantly from theoriginal music score. This inconsistency poses a great challenge in PekingOpera singing voice synthesis from a music score. In this work, we propose todeal with this issue and synthesize expressive Peking Opera singing from themusic score based on the Duration Informed Attention Network (DurIAN)framework. To tackle the rhythm mismatch, Lagrange multiplier is used to findthe optimal output phoneme duration sequence with the constraint of the givennote duration from music score. As for the pitch contour mismatch, instead ofdirectly inferring from music score, we adopt a pseudo music score generatedfrom the real singing and feed it as input during training. The experimentsdemonstrate that with the proposed system we can synthesize Peking Operasinging voice with high-quality timbre, pitch and expressiveness.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...