Table 2. Results of the quality predictions (including anchor conditions) for the normal hearing subjects, expressed by linear correlation coefficients and rank correlation coefficients (italic).
| measure | ind. | av. |
| qc+W-B (M2) | 0.84 | 0.86 |
| 0.89 | 0.93 | |
| qc-W-B (M4) | 0.84 | 0.87 |
| 0.84 | 0.95 | |
| PSM_lp-B (M12) | 0.85 | 0.88 |
| 0.84 | 0.91 | |
| PSM_fb-B (M10) | 0.88 | 0.89 |
| 0.92 | 0.98 | |
| HASQcomb (M19) | 0.54 | 0.54 |
| 0.24 | 0.19 | |
| PESQ (M21) | 0.64 | 0.64 |
| 0.41 | 0.45 | |
| Soveral (M25)l | 0.76 | 0.76 |
| 0.63 | 0.64 | |
| LPD (M22) | −0.75 | −0.77 |
| −0.64 | −0.62 | |
| LAR (M27) | −0.83 | −0.84 |
| −0.74 | −0.71 |
Subjective ratings were averaged across subjects. Middle column: correlations based on ratings for individual talkers (ind.); right column: ratings averaged across talkers (av.). Quality measure suffixes: fb/lp: modulation filterbank/lowpass model version; +/−W: with/without frequency band weighting; +/−B: with/without asymmetric weighting of differences (“Beerends weighting”).