Table 2 Fraction of time frames where the single most useful microphone is identified correctly. The ground-truth selection is given by the microphone with the maximum oracle SNR (9)
From: Microphone utility estimation in acoustic sensor networks using single-channel signal features
Algorithm variant | Synthetic data | Recorded data |
|---|---|---|
average pair-wise CDR [53] | 0.4269 | 0.1717 |
model-KF | 0.5465 | 0.1551 |
hybrid | 0.5698 | 0.1277 |
ML-joint | 0.5793 | 0.1940 |