Abstract | ||
---|---|---|
This paper presents a refinement framework of WaveNet vocoders for variational autoencoder (VAE) based voice conversion (VC), which reduces the quality distortion caused by the mismatch between the training data and testing data. Conventional WaveNet vocoders are trained with natural acoustic features but conditioned on the converted features in the conversion stage for VC, and such a mismatch often causes significant quality and similarity degradation. In this work, we take advantage of the particular structure of VAEs to refine WaveNet vocoders with the self-reconstructed features generated by VAE, which are of similar characteristics with the converted features while having the same temporal structure with the target natural features. We analyze these features and show that the self-reconstructed features are similar to the converted features. Objective and subjective experimental results demonstrate the effectiveness of our proposed framework. |
Year | DOI | Venue |
---|---|---|
2018 | 10.23919/EUSIPCO.2019.8902651 | 2019 27TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO) |
Keywords | Field | DocType |
voice conversion, variational autoencoder, WaveNet vocoder, speaker adaptation | Training set,Autoencoder,Pattern recognition,Computer science,Test data,Artificial intelligence,Distortion | Journal |
Volume | ISSN | Citations |
abs/1811.11078 | 2076-1465 | 3 |
PageRank | References | Authors |
0.37 | 0 | 9 |
Name | Order | Citations | PageRank |
---|---|---|---|
Wen-Chin Huang | 1 | 13 | 7.59 |
Yi-Chiao Wu | 2 | 45 | 9.42 |
Hsin-Te Hwang | 3 | 63 | 7.93 |
Patrick Lumban Tobing | 4 | 15 | 7.89 |
Tomoki Hayashi | 5 | 96 | 18.49 |
Kazuhiro Kobayashi | 6 | 66 | 9.91 |
Tomoki Toda | 7 | 1874 | 167.18 |
Yu Tsao | 8 | 33 | 5.23 |
Hsin-min Wang | 9 | 1201 | 129.62 |