From: Integrated exemplar-based template matching and statistical modeling for continuous speech recognition
Threshold tσ
1σ
2σ
3σ
∞
Accuracy (%)
73.95
74.32
73.42
70.89