notesum.ai

Published at October 20

Anonymising Elderly and Pathological Speech: Voice Conversion Using DDSP and Query-by-Example

cs.CE

cs.NA

math.NA

65M22, 68U20, 76D05

Released Date: October 20, 2024

Authors: Suhita Ghosh¹, Melanie Jouaiti², Arnab Das³, Yamini Sinha⁴, Tim Polzehl³, Ingo Siegert⁴, Sebastian Stober¹

Aff.: ¹Artificial Intelligence Lab, Otto-von-Guericke-University, Magdeburg, Germany; ²School of Computer Science, University of Birmingham; ³Speech and Language Technology, German Research Center for Artificial Intelligence (DFKI); ⁴Mobile Dialog Systems, Otto-von-Guericke-University, Magdeburg, Germany

Arxiv: https://arxiv.org/abs/2410.15500v1

Source

Type

Domain Pr. [%]

\uparrow

PCC [

\times 10^{2}

]

\uparrow

pMOS

\uparrow

CER [%]

\downarrow

EER [%]

\uparrow

Emo

DDSP-QbE

Emo

DDSP-QbE

Emo

DDSP-QbE

Emo

DDSP-QbE

Emo

DDSP-QbE

All conversions

All

67.8

\pm

1.2

78.1

\pm

2.3

68.1

\pm

0.7

77.4

\pm

0.6

2.41

\pm

0.02

3.39

\pm

0.02

18.15

\pm

1.00

1.04

\pm

0.09

50.16

\pm

0.03

48.98

\pm

0.07

Different gender

64.7

\pm

0.9

76.2

\pm

2.5

67.6

\pm

1.0

75.4

\pm

0.9

2.41

\pm

0.03

3.36

\pm

0.03

22.57

\pm

1.41

1.05

\pm

0.12

Same gender

70.9

\pm

1.4

80.0

\pm

2.1

68.6

\pm

0.9

79.4

\pm

0.8

2.42

\pm

0.03

3.41

\pm

0.02

13.73

\pm

1.43

1.02

\pm

0.12

Elderly

\rightarrow

All

61.3

\pm

0.7

79.7

\pm

1.7

55.1

\pm

1.2

70.6

\pm

1.4

2.28

\pm

0.04

3.24

\pm

0.03

24.8

\pm

1.61

2.41

\pm

0.21

40.62

\pm

0.11

43.41

\pm

0.08

Dementia

63.3

\pm

0.9

78.9

\pm

2.5

53.5

\pm

1.7

68.4

\pm

2.0

2.31

\pm

0.05

3.29

\pm

0.04

23.36

\pm

2.26

2.40

\pm

0.29

38.24

\pm

0.08

40.41

\pm

0.05

Healthy

59.3

\pm

0.8

80.5

\pm

0.9

56.6

\pm

1.7

72.8

\pm

1.8

2.24

\pm

0.05

3.20

\pm

0.05

26.24

\pm

2.29

2.42

\pm

0.31

43.40

\pm

0.06

44.39

\pm

0.08

Stuttering

\rightarrow

All

64.3

\pm

2.7

77.8

\pm

2.2

80.0

\pm

0.7

84.2

\pm

0.5

2.41

\pm

0.04

3.33

\pm

0.03

19.07

\pm

4.01

0.13

\pm

0.05

48.99

\pm

0.02

48.85

\pm

0.05

Block

59.7

\pm

3.7

78.3

\pm

2.2

80.3

\pm

1.7

83.8

\pm

1.0

2.41

\pm

0.08

3.37

\pm

0.07

16.55

\pm

4.00

0.08

\pm

0.08

Word Repetition

73.2

\pm

1.7

76.3

\pm

2.1

78.9

\pm

1.6

83.6

\pm

0.9

2.38

\pm

0.09

3.34

\pm

0.06

16.91

\pm

3.93

0.15

\pm

0.10

Sound Repetition

49.8

\pm

4.7

78.0

\pm

2.2

80.3

\pm

1.6

84.6

\pm

1.0

2.37

\pm

0.10

3.22

\pm

0.07

22.50

\pm

3.94

0.20

\pm

0.12

Interjection

74.5

\pm

0.2

78.2

\pm

2.3

79.1

\pm

1.7

83.6

\pm

1.1

2.47

\pm

0.08

3.37

\pm

0.07

20.32

\pm

3.99

0.08

\pm

0.08

\rightarrow

All

77.8

\pm

1.6

76.8

\pm

1.7

79.9

\pm

1.0

81.4

\pm

0.4

3.55

\pm

0.04

3.88

\pm

0.02

10.58

\pm

0.85

0.57

\pm

0.09

52.29

\pm

0.05

51.66

\pm

0.04