Skip to main content

Table 8 Performance comparisons of n-gram, CRF, and auto-context methods

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

 

N-gram[8]

CRF

Auto-context

Accuracy (%)

   

Two-way

80.1

81.1

82.0

Four-way

-

76.9

77.0

SL boundary

89.6

88.2

89.0

WL boundary

84.0

82.2

82.9

F-score (%)

   

Two-way

-

72.5

73.0

Four-way

-

62.9

63.0

SL boundary

-

55.1

57.3

WL boundary

-

56.3

58.9