Table 2 The comparison between the CNN-based model [21], the CRNN-based model, and the base version of the proposed method
From: Exploring the power of pure attention mechanisms in blind room parameter estimation
Method | \(\varvec{\#}\) Params (M) | Evaluation metrics | Memory consumption (GB)) | MACs (G) | |||
|---|---|---|---|---|---|---|---|
MSE | MAE | \(\varvec{\rho}\) | MM | ||||
CNN [21] | 0.013 | 0.3863 | 0.4837 | 0.6984 | 3.0532 | 1.81 | 0.237 |
CRNN | 0.494 | 0.3572 | 0.4265 | 0.7262 | 2.6701 | 1.95 | 0.236 |
Proposed method | 85.256 | 0.2650 | 0.3432 | 0.8077 | 2.2039 | 4.55 | 34.083 |