Skip to main content

Advertisement

Table 3 Statistical comparisons of the SNR LOSS measure between unprocessed sentences and enhanced sentences by SE algorithms

From: A modified Wiener filtering method combined with wavelet thresholding multitaper spectrum for speech enhancement

   WT-UP KLT-UP Wiener_Clean-UP Proposed-UP Proposed-WT Proposed-KLT Proposed-Wiener_Clean
Noise SNR (dB) R(%) pvalue R(%) pvalue R(%) pvalue R(%) pvalue R(%) pvalue R(%) pvalue R(%) pvalue
Train −8 1.30 0.000 0.98 0.011 −7.05 0.000 0.8 7 0.0 3 1 2.1 5 0.000 -1.83 0.000 6.64 0.000
  −5 1.61 0.000 1.26 0.004 −6.87 0.000 −0.91 0.076 2.4 7 0.000 2.1 3 0.000 6.41 0.000
  −2 1.93 0.000 1.43 0.005 −6.74 0.000 −0.96 0.127 2.8 3 0.000 2.3 5 0.000 6.20 0.000
  0 3.41 0.000 2.33 0.002 −9.83 0.000 −1.67 0.054 4.9 2 0.000 3.9 2 0.000 9.04 0.000
  5 3.44 0.001 2.12 0.084 −8.95 0.000 0.45 0.983 2.8 9 0.004 −1.63 0.264 10.32 0.000
  10 2.96 0.024 2.47 0.091 −7.06 0.000 3.74 0.002 0.76 0.930 1.25 0.687 11.63 0.000
  15 3.54 0.034 4.96 0.001 −1.85 0.553 10.76 0.000 6.97 0.000 5.52 0.000 12.85 0.000
Babble −8 1.61 0.000 0.73 0.128 −7.69 0.000 1.1 6 0.002 2.7 3 0.000 1.8 8 0.000 7.07 0.000
  −5 2.25 0.000 1.21 0.010 −7.23 0.000 1.1 1 0.022 3.2 8 0.000 2.3 0 0.000 6.60 0.000
  −2 2.42 0.000 1.22 0.047 −6.90 0.000 −1.06 0.118 3.3 9 0.000 2.2 5 0.000 6.28 0.000
  0 3.25 0.000 1.55 0.046 −9.57 0.000 1.8 5 0.009 4.9 3 0.000 3.3 4 0.000 8.54 0.000
  5 4.88 0.000 2.76 0.002 −8.06 0.000 0.10 1.000 4.5 6 0.000 2.5 9 0.004 8.87 0.000
  10 6.11 0.000 4.74 0.000 −4.76 0.000 4.72 0.000 −1.31 0.684 −0.02 1.000 9.95 0.000
  15 6.84 0.000 7.72 0.000 1.91 0.687 12.76 0.000 5.54 0.001 4.67 0.006 10.65 0.000
Car −8 1.18 0.000 0.12 0.984 −8.38 0.000 1.8 9 0.000 3.0 4 0.000 2.0 1 0.000 7.09 0.000
  −5 1.90 0.000 0.46 0.424 −8.20 0.000 2.1 1 0.000 3.9 4 0.000 2.5 6 0.000 6.64 0.000
  −2 2.25 0.000 0.56 0.321 −8.20 0.000 2.4 8 0.000 4.6 2 0.000 3.0 3 0.000 6.23 0.000
  0 3.38 0.000 0.78 0.373 −11.23 0.000 3.4 5 0.000 6.6 1 0.000 4.1 9 0.000 8.77 0.000
  5 3.46 0.000 −0.21 0.997 −10.70 0.000 2.3 3 0.001 5.5 9 0.000 2.1 2 0.005 9.38 0.000
  10 5.32 0.000 1.90 0.345 −7.52 0.000 1.55 0.554 3.5 8 0.003 −0.34 0.997 9.81 0.000
  15 5.10 0.000 2.78 0.075 −2.68 0.093 8.25 0.000 2.99 0.030 5.32 0.000 11.22 0.000
Exhibition Hall −8 0.58 0.026 −0.42 0.201 −7.79 0.000 1.4 4 0.000 2.0 1 0.000 1.0 3 0.000 6.88 0.000
  −5 1.01 0.000 −0.17 0.947 −7.70 0.000 1.6 1 0.000 2.6 0 0.000 1.4 5 0.000 6.60 0.000
  −2 1.41 0.000 0.03 1.000 −7.48 0.000 1.8 3 0.000 3.2 0 0.000 1.8 6 0.000 6.10 0.000
  0 4.00 0.000 0.77 0.738 −10.52 0.000 1.9 5 0.020 5.7 2 0.000 2.7 0 0.000 9.58 0.000
  5 3.68 0.000 −0.62 0.889 −9.69 0.000 0.15 0.999 3.4 0 0.000 0.77 0.779 10.90 0.000
  10 4.86 0.001 2.14 0.393 −6.57 0.000 4.34 0.004 −0.49 0.993 2.15 0.365 11.67 0.000
  15 6.17 0.000 4.88 0.003 −1.24 0.880 11.13 0.000 4.67 0.002 5.96 0.000 12.53 0.000
Restaurant −8 1.81 0.000 0.95 0.033 −7.42 0.000 1.1 4 0.005 2.9 0 0.000 2.0 7 0.000 6.78 0.000
  −5 2.15 0.000 1.10 0.044 −7.09 0.000 −1.07 0.053 3.1 6 0.000 2.1 5 0.000 6.47 0.000
  −2 2.18 0.000 1.12 0.113 −6.67 0.000 −1.05 0.159 3.1 6 0.000 2.1 4 0.000 6.03 0.000
  0 4.76 0.000 3.21 0.000 −8.82 0.000 −0.88 0.746 5.3 8 0.000 3.9 6 0.000 8.70 0.000
  5 4.06 0.002 2.53 0.142 −7.90 0.000 0.83 0.939 3.1 0 0.028 −1.65 0.525 9.49 0.000
  10 7.40 0.000 6.27 0.000 −3.21 0.053 6.53 0.000 −0.81 0.945 0.24 1.000 10.06 0.000
  15 8.27 0.000 9.01 0.000 −3.77 0.101 15.18 0.000 6.38 0.000 5.67 0.001 11.00 0.000
Street −8 1.11 0.045 0.72 0.363 −7.26 0.000 1.2 7 0.015 2.3 5 0.000 1.9 8 0.000 6.46 0.000
  −5 1.48 0.017 0.89 0.329 −7.24 0.000 1.3 2 0.044 2.7 6 0.000 2.1 9 0.000 6.39 0.000
  −2 1.58 0.030 0.86 0.502 −7.12 0.000 −1.36 0.088 2.9 0 0.000 2.2 0 0.001 6.21 0.000
  0 4.76 0.000 2.90 0.004 −9.14 0.000 −1.13 0.622 5.6 2 0.000 3.9 1 0.000 8.82 0.000
  5 4.64 0.000 2.85 0.036 −8.59 0.000 0.74 0.944 3.7 3 0.001 −2.05 0.210 10.20 0.000
  10 6.28 0.001 5.29 0.007 −5.61 0.003 5.36 0.006 −0.86 0.976 0.07 1.000 11.62 0.000
  15 7.55 0.000 7.49 0.000 0.47 0.998 12.22 0.000 4.34 0.014 4.41 0.012 11.70 0.000
Airport −8 1.80 0.000 0.96 0.147 −7.89 0.000 1.5 1 0.004 3.2 5 0.000 2.4 4 0.000 6.93 0.000
  −5 2.31 0.000 1.37 0.048 −7.42 0.000 1.3 7 0.048 3.6 0 0.000 2.7 0 0.000 6.53 0.000
  −2 2.65 0.000 1.60 0.035 −6.78 0.000 −1.13 0.249 3.6 8 0.000 2.6 8 0.000 6.06 0.000
  0 4.53 0.000 1.72 0.225 −9.43 0.000 −1.45 0.392 5.7 2 0.000 3.1 2 0.001 8.81 0.000
  5 5.26 0.000 3.36 0.007 −7.69 0.000 0.60 0.972 4.4 2 0.000 2.6 7 0.042 8.99 0.000
  10 7.72 0.000 5.88 0.000 −2.90 0.113 6.58 0.000 −1.05 0.876 0.66 0.977 9.76 0.000
  15 7.76 0.000 7.43 0.000 2.61 0.270 13.81 0.000 5.62 0.000 5.95 0.000 10.92 0.000
Train station −8 1.67 0.000 0.89 0.034 −8.37 0.000 1.9 1 0.000 3.5 2 0.000 2.7 8 0.000 7.04 0.000
  −5 2.30 0.000 1.22 0.010 −7.89 0.000 1.9 0 0.000 4.1 1 0.000 3.0 9 0.000 6.50 0.000
  −2 2.64 0.000 1.47 0.006 −7.53 0.000 1.8 8 0.000 4.4 1 0.000 3.3 0 0.000 6.11 0.000
  0 3.19 0.000 0.58 0.857 −10.72 0.000 2.9 7 0.000 5.9 8 0.000 3.5 4 0.000 8.67 0.000
  5 4.25 0.003 1.24 0.812 −9.47 0.000 −1.63 0.608 5.6 4 0.000 −2.84 0.092 8.65 0.000
  10 5.43 0.000 2.38 0.192 −5.94 0.000 2.79 0.082 −2.50 0.115 0.41 0.995 9.29 0.082
  15 7.60 0.000 6.30 0.000 0.38 0.999 11.19 0.000 3.33 0.065 4.60 0.004 10.76 0.000
  1. Four SE algorithms: WT, KLT, Wiener_Clean, and proposed. The comparisons between our proposed SE algorithm and the other three were also given. The results in italics show the SNRLOSS measure by which our proposed SE algorithm is better than others, and the difference was significant (p <0.05).