Table 9 Subjective comparison results with standard deviation (std dev) of mono-lingual conversion in unseen-to-unseen scenario. The results are listd as MOS/std dev
From: U2-VC: one-shot voice conversion using two-level nested U-structure
| Â | MOS (similarity)/std dev | MOS (naturalness)/std dev | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Â | SF2TF | SF2TM | SM2TF | SM2TM | Average | SF2TF | SF2TM | SM2TF | SM2TM | Average |
AdaIN-VC | 2.01/0.31 | 2.02/0.48 | 2.04/0.29 | 2.10/0.29 | 2.04 | 2.07/0.61 | 2.08/0.67 | 2.05/0.58 | 2.06/0.68 | 2.07 |
AGAIN-VC | 2.68/0.49 | 2.88/0.57 | 2.94/0.47 | 3.31/0.35 | 2.95 | 3.14/0.82 | 3.24/0.65 | 3.02/0.82 | 3.50/0.51 | 3.23 |
U2-VC | 3.14/0.43 | 3.40/0.49 | 3.34/0.43 | 3.62/0.37 | 3.37 | 3.94/0.77 | 4.04/0.47 | 3.74/0.78 | 4.05/0.56 | 3.94 |