Table 2.

Performance comparison with baselines.

Method	DTW/1.636e-3	PSD/1.128e-9	HD/2.886e-6
ASAE (Antoniades et al., 2018)	2.867 ± 0.126	−1.384 ± 1.794	0.490 ± 0.800
AE(*)	−0.059 ± 0.815	−1.357 ± 1.659	0.337 ± 0.756
EEGGAN (Hartmann et al., 2018)	−0.019 ± 0.624	−0.821 ± 1.457	0.068 ± 0.734
uTSGAN (Smith and Smith, 2021)	3.867 ± 0.012	3.424 ± 0.134	3.127 ± 0.050
Pix2pix (Isola et al., 2017)	0.133 ± 0.829	−1.396 ± 1.754	0.316 ± 0.834
GANSynth (Engel et al., 2019)	−0.107 ± 0.740	−0.929 ± 1.453	0.191 ± 0.763
E2SGAN-N	−0.424 ± 0.400	0.185 ± 0.298	0.572 ± 0.565
E2SGAN-S	−0.414 ± 0.764	−1.480 ± 1.609	−0.221 ± 0.843

AE^(*) is implemented with the same architecture as the generator of the proposed method. The bold values indicate the baseline values of evaluation.