EVA-GAN: Enhanced Various Audio Generation via Scalable Generative Adversarial Networks

Benchmark Model Rank Results
speech-synthesis-on-librittsEVA-GAN-big#3PESQ: 4.3536M-STFT: 0.7982Periodicity: 0.0751V/UV F1: 0.9745
speech-synthesis-on-librittsEVA-GAN-base#8PESQ: 4.0330M-STFT: 0.9485Periodicity: 0.0942V/UV F1: 0.9658