Skip to main content

Table 4 Different frameworks’ UAR results for classification subtasks of three 4-class paralinguistic singing attributes

From: Paralinguistic singing attribute recognition using supervised machine learning for describing the classical tenor solo singing voice in vocal pedagogy

 

Unweighted average recall (UAR) [%]

Frameworks

Chest resonance

Head resonance

Open throat

Average

SF-SVM (ComparE09)

34

37.21

28.1

33.10

SF-SVM (ComparE016)

38.7

34.34

46.2

39.74

E2EDL (ResNet)

44.39

37.33

28.8

36.84

E2EDL (Transformer)

41.54

37.68

29

36.07

DE-SVM (ResNet)

46.63

44.43

30.82

40.63

DE-SVM (Transformer)

42.17

40.26

22.58

35