Table 5 Medians for newGASpeech (GA) and WinSnoori (Wsno) by voice type for male speakers
From: Utterance copy through analysis-by-synthesis using genetic algorithm
Male | Female | |||||||
|---|---|---|---|---|---|---|---|---|
Natural voice | Synthesized voice | Natural voice | Synthesized voice | |||||
Metrics/median | GA | Wsno | GA | Wsno | GA | Wsno | GA | Wsno |
RMSE | 50.3 | 126.9 | 78.3 | 585.6 | 54.8 | 221.2 | 97 | 799 |
D LE | 0.04 | 0.38 | 0.004 | 0.05 | 0.028 | 0.26 | 0.004 | 0.036 |
PESQ | 2.5 | 1.7 | 3.4 | 1.5 | 2.9 | 2.0 | 3.7 | 2.3 |
P.563 | – | – | 4.4 | 1.5 | – | – | 4.5 | 1 |
SNR | 8.6 | 0.3 | 4.2 | −5.2 | 15.1 | 1.2 | 131.4 | 0.9 |