Table 2 Performance comparison - correlation between the (temporal mean) proposed power ratio-based DRR estimator (PR-DRR) (19) and Jeub et al. CDR-based DRR estimator, with the true DRR measure
From: Microphone array power ratio for quality assessment of reverberated speech
Input type | White noise | Speech signals | |
|---|---|---|---|
Correlation ref. | Correlation ref. | ||
Test type | Algorithm | DRR | DRR |
T 60=0.3 s, | PR-DRR | 0.999 | 0.999 |
vs. increasing distance | CDR | 0.995 | 0.992 |
T 60=1 s, | PR-DRR | 0.999 | 0.999 |
vs. increasing distance | CDR | 0.964 | 0.972 |
Distance =0.5 m, | PR-DRR | 0.994 | 0.996 |
vs. increasing T 60 | CDR | 0.984 | 0.978 |
Distance =2 m, | PR-DRR | 0.999 | 0.999 |
vs. increasing T 60 | CDR | 0.852 | 0.913 |