eduzhai > Applied Sciences > Engineering >

Voice Conversion Challenge 2020 Intra-lingual semi-parallel and cross-lingual voice conversion

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 19 pages

Abstract: The voice conversion challenge is a bi-annual scientific event held tocompare and understand different voice conversion (VC) systems built on acommon dataset. In 2020, we organized the third edition of the challenge andconstructed and distributed a new database for two tasks, intra-lingualsemi-parallel and cross-lingual VC. After a two-month challenge period, wereceived 33 submissions, including 3 baselines built on the database. From theresults of crowd-sourced listening tests, we observed that VC methods haveprogressed rapidly thanks to advanced deep learning methods. In particular,speaker similarity scores of several systems turned out to be as high as targetspeakers in the intra-lingual semi-parallel VC task. However, we confirmed thatnone of them have achieved human-level naturalness yet for the same task. Thecross-lingual conversion task is, as expected, a more difficult task, and theoverall naturalness and similarity scores were lower than those for theintra-lingual conversion task. However, we observed encouraging results, andthe MOS scores of the best systems were higher than 4.0. We also show a fewadditional analysis results to aid in understanding cross-lingual VC better.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...