Skip to main content

Table 3 The average segSNRs with and without postfiltering for the typical DNN-based speech enhancement methods

From: Low-complexity artificial noise suppression methods for deep learning-based speech enhancement algorithms

SNR/dB − 5 0 5 10 Aver. SNR/dB − 5 0 5 10 Aver.
Noisy −7.00 −4.06 −0.85 2.69 −2.31 Noisy −7.00 −4.06 −0.85 2.69 −2.31
CRN −0.03 3.61 5.62 9.18 4.60 DCN −0.28 3.25 5.68 8.82 4.37
SPP-MMSE 0.46 4.10 6.14 9.71 5.10 SPP-MMSE 0.26 3.65 6.14 9.40 4.86
SPP-proposed-1 1.34 4.50 6.75 9.95 5.64 SPP-proposed-1 1.34 4.28 6.79 9.70 5.53
SPP-proposed-2 1.27 4.38 6.60 9.76 5.50 SPP-proposed-2 1.30 4.15 6.59 9.50 5.39
SPP-proposed-3 1.25 4.62 6.68 10.05 5.65 SPP-proposed-3 1.32 4.29 6.64 9.76 5.50
GRN 0.11 3.99 6.02 9.28 4.85 DARCN 0.67 3.99 6.13 9.10 4.97
SPP-MMSE 0.57 4.28 6.33 9.62 5.20 SPP-MMSE 1.10 4.46 6.48 9.37 5.35
SPP-proposed-1 1.66 4.79 6.97 9.90 5.83 SPP-proposed-1 1.73 4.66 6.69 9.40 5.62
SPP-proposed-2 1.53 4.67 6.80 9.76 5.69 SPP-proposed-2 1.35 3.98 5.87 8.34 4.89
SPP-proposed-3 1.57 4.91 6.87 9.93 5.82 SPP-proposed-3 1.73 4.85 6.75 9.60 5.73