Skip to main content

Table 4 Mean objective results for pink noise

From: An adaptive a priori SNR estimator for perceptual speech enhancement

Gain SNR PESQ SNR seg HD KurtR
   DD MDD TSNR Prop DD MDD TSNR Prop DD MDD TSNR Prop DD MDD TSNR Prop
WF 0 1.887 1.874 1.879 1 . 9 6 5 0.253 0.257 0.057 0 . 6 6 2 0.405 0.417 0.407 0 . 4 0 3 1.030 1.014 1 . 0 1 3 1.026
  5 2.355 2.344 2.327 2 . 3 9 9 2.671 2.710 2.948 3 . 3 7 9 0.389 0.394 0.399 0 . 3 8 1 1.173 1.067 1 . 0 5 7 1.102
  10 2.735 2.725 2.696 2 . 7 8 2 5.616 5.683 6.037 6 . 3 8 8 0.364 0.372 0.375 0 . 3 2 6 1.554 1.231 1 . 2 0 8 1.339
LSA 0 1.978 1.969 1.768 2 . 0 4 3 0.184 0.364 −0.399 0 . 3 8 7 0.399 0.397 0.395 0 . 3 9 0 1.245 1 . 0 5 2 3.724 1.077
  5 2.405 2.437 2.266 2 . 4 9 7 2.608 2.899 2.532 3 . 3 9 6 0 . 3 6 0 0.381 0.369 0.361 1.795 1 . 1 4 1 3.799 1.200
  10 2.762 2.805 2.670 2 . 8 6 9 5.637 5.966 5.837 6 . 7 2 0 0 . 3 2 0 0.354 0.350 0.343 2.708 1 . 3 6 7 3.941 1.512
  1. Bold values denote the best performance