Skip to main content

Table 7 Mean objective results for babble 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 1.934 1.931 1.922 1 . 9 3 6 −0.325 −0.305 −0.650 0 . 2 6 6 0 . 3 9 8 0.401 0.402 0.400 1.257 1.181 1 . 1 7 1 1.192
  5 2.280 2.285 2.273 2 . 2 8 5 1.628 1.679 1.795 1 . 8 8 7 0 . 3 7 4 0.381 0.382 0.379 1.509 1.273 1 . 2 3 2 1.293
  10 2.598 2.605 2.593 2 . 6 7 0 4.350 4.459 4.664 4 . 6 8 7 0 . 3 3 8 0.350 0.351 0.347 2.056 1.576 1 . 4 7 9 1.612
LSA 0 1.966 1.979 1.842 1 . 9 8 1 −0.426 −0.273 −0.963 0 . 2 4 3 0 . 3 6 1 0.383 0.382 0.380 1.620 1 . 4 0 8 2.205 1.423
  5 2.290 2.332 2.199 2 . 3 3 5 1.528 1.788 1.490 1 . 8 7 0 0 . 3 2 6 0.358 0.360 0.357 2.041 1 . 5 3 9 2.252 1.556
  10 2.611 2.673 2.561 2 . 6 7 4 4.288 4.636 4.473 4 . 7 9 6 0 . 2 8 1 0.323 0.322 0.321 2.504 1 . 8 1 9 2.575 1.849
  1. Bold values denote the best performance