Skip to main content

Table 1 Summary of different approaches for prosodic event detection

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

Reference

Corpus

Domain

Model

Class

Pitch accent

Boundary

     

accuracy (%)

accuracy (%)

[4]

TTS and BN

Word

Prosodic context-dependent

Binary

82.8

-

   

HMM

   

[5]

BURSC

Word, syllable

CHMM

Binary

72.0

77.3

     

74.0

86.0

[14]

BURSC

Syllable

GMM

Binary

77.3

68.2

[7]

BURSC

Syllable

Decision tree and

Binary

84.0

71.0

   

n-gram language model

   

[8]

BURSC

Syllable

NN and

Binary

80.1

89.6

   

n-gram language model

   

[9]

BURSC

Syllable

NN

Binary

83.5

84.8

[10]

BURSC

Syllable

SVM

Binary

85.7

-

[6]

BURSC

Syllable

Adaboost CART

Four

84.7

-

[15]

BURSC

Word

MaxEnt

Binary

80.1

82.7

[11]

BURSC

Syllable

CRF

Binary, four

79.5 77.1

92.4