Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing

Table 4 Performance of prosodic event detection by NN and SVM classifiers using all combined features

From: Exploiting contextual information for prosodic event detection using auto-context

 

Two-way (%)

Four-way (%)

SL boundary (%)

WL boundary (%)

 

Accuracy

F-score

Accuracy

F-score

Accuracy

F-score

Accuracy

F-score

NN

80.5

66.8

75.0

58.9

87.7

43.7

80.3

44.1

SVM

81.2

70.8

76.3

59.9

88.1

45.3

81.2

46.2

  1. Here ‘Two-way’ and ‘Four-way’ mean the pitch accent detection tasks in two-way and four-way modes, respectively, and ‘SL boundary’ and ‘WL boundary’ mean the boundary detection tasks measured at the syllable-level and word-level, respectively.
Back to article page