Supplementary Material of HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis Appendix A

The detailed architecture of the generator and MPD is depicted in Figure 4. Therefore, V3 consists of a much smaller number of layers than V1 and V2. 13 Appendix B We gave true label [99, 99.5, 99.9]% of the We repeated this experiment 5 times to get the average, and the results are listed in Table 6. The results show that MPD is superior in discriminating periodic signals than MSD. Figure 5b shows input signals of sub-discriminators and the magnitude of their frequency responses. In the case of MPD, the frequency responses of input signals are not distorted except for aliasing. On the other hand, the input signals of MSD are getting smoother whenever down-sampling. When comparing the outputs of learned generators, the difference is more evident.