Abstract | ||
---|---|---|
•The proposed voice conversion pipeline, DeepConversion, leverages a large amount of non-parallel data, but requires only a small amount of parallel training data.•We propose a strategy to make full use of the parallel data in all models along the pipeline.•The parallel data is also used to adapt the WaveNet vocoder towards the source-target pair.•The experiments show that DeepConversion outperforms the traditional approaches in both objective and subjective evaluations. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1016/j.specom.2020.05.004 | Speech Communication |
Keywords | DocType | Volume |
Voice conversion,Limited data,Deep learning,Wavenet | Journal | 122 |
ISSN | Citations | PageRank |
0167-6393 | 3 | 0.36 |
References | Authors | |
54 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Mingyang Zhang | 1 | 104 | 10.61 |
Berrak Sisman | 2 | 60 | 10.34 |
Li Zhao | 3 | 198 | 22.70 |
Haizhou Li | 4 | 3678 | 334.61 |