Table 10 Comparison of parameter counts and multiply-accumulate operations. Here, ✓ indicates causal models
From: Supervised Attention Multi-Scale Temporal Convolutional Network for monaural speech enhancement
Causal | #Param. (M) | MACs (G/s) | |
|---|---|---|---|
CRN | ✓ | 17.58 | 2.54 |
MSTCN | ✓ | 4.78 | 0.48 |
LSTM-IRM | ✓ | 13.24 | 1.34 |
GCRN | ✓ | 9.77 | 2.47 |
GaGNet | ✓ | 5.94 | 1.63 |
Conv-TasNet | ✓ | 5.00 | 5.23 |
DCCRN | ✓ | 3.67 | 14.38 |
DPCRN | ✓ | 0.80 | 3.17 |
SA-MSTCN\(^{1}\) | ✓ | 24.66 | 8.45 |
SA-MSTCN\(^{2}\) | ✓ | 28.12 | 12.07 |