Skip to main content

Table 1 The standard of speech quality evaluation PESQ

From: Improved Wasserstein conditional generative adversarial network speech enhancement

Scenes PESQ 2 3 4 6 8 Average
IWCGAN 1.915 1.714 1.497 1.714 1.634 1.695
SEGAN 1.908 1.443 1.423 1.412 1.855 1.608
  1. Improved model achieves a higher average PESQ score than that by the SEGAN model, which means added elastic work improves the intelligibility and speech quality. The PESQ improvements from 5 to 21% are observed