21.3. Results of the statistical analysis

The patterns of data dependency between a parametric description of the EGG signal and an expert voice classification were also the subject of a more classical statistical data analysis. The results of ANOVA for groups of speakers are summarized in Table 16. The grouping of speakers was found to be independent of vowel quality. The results of each of the measured parameters are given relative to the classification of the speakers.

Table 16: The mean and standard deviation (given in parentheses) of the parameters for the respective subject group. The significant differences (alpha=0.05) are bold.

Parameter

control

breathy

rp-with-comp

rp-without-c

chordectomy

Perturbation of F0

-0.018 (0.022)

0.022 (0.142)

-0.235 (0.677)

0.063 (0.593)

-1.142 (2.988)

Open Quotient I [%]

64.7 (3.56)

62.6 (5.36)

66.5 (7.8)

60.1 (7.08)

59.1 (12.3)

Open Duration [%]

18.3 (2.4)

26.78 (7.8)

23.38 (11.3)

18.2 (6.8)

9.7 (8.7)

Speed Quotient [%]

76.2 (23.2)

80.8 (37.5)

130.0 (57.07)

215.8 (362.9)

143.4 (60.4)

Peak-to-peak

31094 (8109)

29570 (7057)

24805 (9300)

21919 (9697)

19113 (11732)

Closing Slope

1146 (229)

1317 (691)

915 (568)

288 (143)

234 (127)

Opening Slope

-242.6 (111)

-294 (138.5)

-302.8 (179)

-187 (100)

-235.4 (81.1)

Opening Duration [%]

20.47 (1.8)

20.4 (4.1)

15.9 (4.4)

14.9 (4.2)

11.1 (3.8)

Closing Duration [%]

5.1 (2.1)

6.1 (3.4)

6.9 (2.5)

11.9 (4.1)

13.2 (3.9)

Closed Duration [%]

10.4 (1.1)

11.6 (2.4)

10.6 (3)

12.5 (3.1)

12.1 (3.5)

variation of Closed Phase

1767 (453)

1825.3 (425)

1489 (600)

1227 (664)

1105.3 (575)

variation of Open Phase

1693 (485)

1665 (355)

1353 (545)

1209 (611)

1065 (687)