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

Table 7 Results for the Albayzin evaluation winner system and factor analysis segmentation system over the test files

From: Audio segmentation-by-classification approach based on factor analysis in broadcast news domain

 

Error for each class

  
 

MU

SP

SM

SN

Total

NIST

HMM-Winn [31]

19.2

39.5

25.0

37.2

30.2

-

HMM-Rep

16.3

40.8

24.0

38.8

30.0

19.3

Worst FA-Segm

19.3

29.5

24.6

33.1

26.6

16.7

Best FA-Segm

18.8

23.7

23.6

29.1

23.8

14.7

  1. The table shows the error per class and the total error with the metric of the evaluation and the NIST metric. The italicized numbers show the total error and the NIST metric of the best FA system.
Back to article page