Table 2. Paired sample t-test and Cohen’s D for TorchDIVA and DiffWave speech quality metrics.
| Group 1 M (SD) | Group 2 M (SD) | t_stat | p_val | cohens_D | |
|---|---|---|---|---|---|
| PESQ | 1.06 (0.053) | 1.622 (0.61) | -2.772 | 1.08E-02 | -1.296 |
| CSIG | -5.543 (1.319) | 2.841 (1.497) | -15.285 | 4.79E-08 | -5.942 |
| CBAK | -0.372 (0.379) | 1.566 (0.778) | -7.024 | 3.08E-05 | -3.168 |
| COVL | -2.974 (0.712) | 2.111 (1.139) | -13.912 | 1.08E-07 | -5.351 |
| SEGSNR | -4.347 (3.471) | -5.573 (3.147) | 3.152 | 5.85E-03 | 0.370 |
Paired sample t-test results for all speech quality metrics. Group 1 represents TorchDIVA output with no enhancement, while Group 2 represents the DiffWave-enhanced speech. The final column of the table contains the Cohen’s D (effect size).