Table 3 Comparison of the GNSDR, GSIR, and GSAR scores of the proposed RPCA-DRNN method with RPCA_b, RPCA_p, sRNN, MLRR, RNMF, MOD-GD, U-Net, EFN, and CRNN-A on MIR-1K
From: RPCA-DRNN technique for monaural singing voice separation
Vocal | |||
Model method | GNSDR | GSIR | GSAR |
 RPCA_b | 2.72 | 3.10 | 5.90 |
 RPCA_p | 4.76 | 4.44 | 8.19 |
 sRNN | 6.43 | 7.69 | 6.82 |
 MLRR | 3.85 | 5.63 | 10.70 |
 RNMF | 4.97 | 7.66 | 10.03 |
 MOD-GD | 7.50 | 13.73 | 9.45 |
 U-Net | 7.43 | 11.79 | 10.42 |
 EFN | 7.76 | 12.97 | 10.16 |
 CRNN-A | 7.89 | 13.75 | 10.17 |
 RPCA-DRNN | 8.46 | 7.72 | 10.83 |
Accompaniment | |||
Model method | GNSDR | GSIR | GSAR |
 RPCA_b | 2.87 | 10.02 | 4.88 |
 RPCA_p | 4.91 | 14.03 | 5.80 |
 sRNN | 6.20 | 8.76 | 6.46 |
 MLRR | 4.19 | 7.80 | 8.22 |
 RNMF | - | - | - |
 MOD-GD | - | - | - |
 U-Net | - | - | - |
 EFN | 7.86 | 13.54 | 9.87 |
 CRNN-A | 7.12 | 9.62 | 11.97 |
 RPCA-DRNN | 8.02 | 12.32 | 11.99 |