Automatic Detection of Parkinsons Disease from Continuous Speech - - PowerPoint PPT Presentation

▶
automatic detection of parkinson s disease
SMART_READER_LITE
LIVE PREVIEW

Automatic Detection of Parkinsons Disease from Continuous Speech - - PowerPoint PPT Presentation

Dec 01, 2023 232 likes •588 views

Automatic Detection of Parkinsons Disease from Continuous Speech Recorded in Non- Controlled Noise Conditions J.C. Vsquez-Correa 1 , T. Arias-Vergara 1 , J.R Orozco-Arroyave 1,2 , J.F Vargas-Bonilla 1 , J.D Arias-Londoo 1 , Elmar Nth 2 1

slide-1
SLIDE 1

Automatic Detection of Parkinson’s Disease from Continuous Speech Recorded in Non- Controlled Noise Conditions

J.C. Vásquez-Correa1, T. Arias-Vergara1, J.R Orozco-Arroyave1,2, J.F Vargas-Bonilla1, J.D Arias-Londoño1, Elmar Nöth2

1 Faculty of Engineering, University of Antioquia UdeA 2 Pattern Recognition Lab, Friedrich Alexander Universität, Erlangen-Nürnberg

Interspeech 2015

slide-2
SLIDE 2

2

  • 1. Introduction
  • 2. Methodology
  • 3. Database
  • 4. Device
  • 5. Results
  • 6. Conclusion

Outline

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-3
SLIDE 3
  • 1. Introduction

3

Missarticulation Monotonic speech Reduced Loudness

 Voice impairments appear in about 90%

  • f

people with Parkinson’s diasease.  Only from 3% to 4% of patients recieve speech therapy

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-4
SLIDE 4

4

  • 1. Introduction

 Intereset: Develop computer-aided tools to perform screenings from

  • speech. The main aims are

 Spare patients moving from home to the hospital  Raise early alerts to patient and doctors

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-5
SLIDE 5

5

  • 1. Introduction

interspeech-2015

jcamilo.vasquez@udea.edu.co

Currently

 The devices are mainly focused on the analysis of sustained phonations.  The devices could be considered invasives.  Methodologies are not adapted for evaluating speech recorded in non- controlled noise conditions.

slide-6
SLIDE 6

6

  • 1. Introduction

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-7
SLIDE 7

7

  • 1. Introduction

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-8
SLIDE 8

8

  • 1. Introduction

 Portable devices and tools for the analysis of speech of people wirth Parkinson’s disease considering continuous speech signals recorded in non-controlled noise conditions.

interspeech-2015

jcamilo.vasquez@udea.edu.co

What is missing?

slide-9
SLIDE 9

9

  • 1. Introduction

 Portable device for recording and analysis of speech of patients  Evaluation of methodology in non-controlled noise conditions.

interspeech-2015

jcamilo.vasquez@udea.edu.co

Aims

slide-10
SLIDE 10

10

  • 1. Introduction
  • 2. Methodology
  • 3. Databases
  • 4. Device
  • 5. Results
  • 6. Conclusion

Outline

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-11
SLIDE 11

11

  • 2. Methodology

Speech signal Pre- processing V/U segmentation

Voiced

Unvoiced ΔF0, Jitter, Shimmer, log-Energy, MFCC MFCC BBE

SVM SVM

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-12
SLIDE 12

12

  • 2. Methodology

Speech signal Pre- processing V/U segmentation

Voiced

Unvoiced ΔF0, Jitter, Shimmer, log-Energy, MFCC MFCC BBE

SVM SVM

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-13
SLIDE 13

13

  • 2. Methodology

Pre-processing

 Speech enhancement  Mean cepstral subtraction

interspeech-2015

jcamilo.vasquez@udea.edu.co

0.5 1 1.5 2

  • 1
  • 0,5

1 Noisy Signal 0.5 1 1.5 2

  • 1

0,1 Time [s] Enhanced Signal 0.5 1 1.5 2 1 2 x 10

4

Noisy Signal Frequency (Hz) 0.5 1 1.5 2 1 2 x 10

4

Time [s] Enhanced Signal Frequency (Hz)

slide-14
SLIDE 14

14

  • 2. Methodology

Speech signal Pre- processing V/U segmentation

Voiced

Unvoiced ΔF0, Jitter, Shimmer, log-Energy, MFCC MFCC BBE

SVM SVM

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-15
SLIDE 15

15

Two types of sound:  Voiced  Unvoiced Both kind

  • f

segments are processed independently

  • 2. Methodology

Segmentation interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-16
SLIDE 16

16

  • 2. Methodology

Speech signal Pre- processing V/U segmentation

Voiced

Unvoiced ΔF0, Jitter, Shimmer, log-Energy, MFCC MFCC BBE

SVM SVM

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-17
SLIDE 17

17

  • 2. Methodology

Characterization

 ΔF0  Jitter  Shimmer  log-Energy  12 MFCC  12 MFCC  25 Energy coefficients acccording to Bark scale

Features Voiced Segments Features Unvoiced Segments interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-18
SLIDE 18

18

  • 2. Methodology

Characterization interspeech-2015

jcamilo.vasquez@udea.edu.co

mean, standard deviation, skewness, kurtosis

2 4 6 8 10 12 x 104

  • 0.8
  • 0.6
  • 0.4
  • 0.2

0.2 0.4 0.6

500 1000 1500
  • 0.4
  • 0.3
  • 0.2
  • 0.1
0.1 0.2 0.3 0.4

20 ms length Time-shift of 10 ms

slide-19
SLIDE 19

19

  • 2. Methodology

Speech signal Pre- processing V/U segmentation

Voiced

Unvoiced ΔF0, Jitter, Shimmer, log-Energy, MFCC MFCC BBE

SVM SVM

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-20
SLIDE 20

20

Classification

  • 2. Methodology

 Gaussian kernel SVM.  Parameters of the SVM are optimized in a range:  10-1 < C < 10 4  10-2 < γ < 10 2  Leave One Speaker Out (LOSO) Cross-validation

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-21
SLIDE 21

21

  • 1. Introduction
  • 2. Methodology
  • 3. Database
  • 4. Device
  • 5. Results
  • 6. Conclusion

Outline

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-22
SLIDE 22

22

  • 3. Database

Patients with Parkinson’s disease Healthy controls Num recordings 14 14 Age Mean 61.64 ± 6.43 Mean 63.29 ± 10.43 Gender 7 male, 7 female 7 male, 7 female Sampling frequency 44100 Hz Quantization bits 16

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-23
SLIDE 23

23

  • 3. Database

 Six different sentences  One read text with 36 words

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-24
SLIDE 24

24

  • 1. Introduction
  • 2. Methodology
  • 3. Database
  • 4. Device
  • 5. Results
  • 6. Conclusion

Outline

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-25
SLIDE 25

25

  • 3. Database

Recording Device

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-26
SLIDE 26

26

  • 1. Introduction
  • 2. Methodology
  • 3. Database
  • 4. Device
  • 5. Results
  • 6. Conclusion

Outline

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-27
SLIDE 27

27

jcamilo.vasquez@udea.edu.co

  • 4. Results

Voiced segments

Sentence Accuracy (%) Noisy Enhanced 1 71 ± 26 82 ± 25 2 75 ± 26 64 ± 36 3 71 ± 26 79 ± 25 4 86 ± 31 79 ± 25 5 79 ± 25 82 ± 25 6 86 ± 23 75 ± 26 Read text 79 ± 25 71 ± 26

interspeech-2015

slide-28
SLIDE 28

28

Sentence Accuracy (%) Noisy Enhanced 1 92 ± 19 93 ± 17 2 94 ± 15 91 ± 20 3 86 ± 23 97 ± 12 4 93 ± 18 94 ± 16 5 78 ± 25 90 ± 20 6 86 ± 23 97 ± 12 Read text 99 ± 3 99 ± 1

  • 4. Results

Unvoiced segments interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-29
SLIDE 29

29

Sentence Accuracy (%) Noisy Enhanced 1 92 ± 19 93 ± 17 2 94 ± 15 91 ± 20 3 86 ± 23 97 ± 12 4 93 ± 18 94 ± 16 5 78 ± 25 90 ± 20 6 86 ± 23 97 ± 12 Read text 99 ± 3 99 ± 1

  • 4. Results

Unvoiced segments interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-30
SLIDE 30

30

0.5 1 0.2 0.4 0.6 0.8 1 False Positive Rate True Positive Rate

  • 4. Results

0.5 1 0.2 0.4 0.6 0.8 1 False Positive Rate True Positive Rate Enhanced Noisy

Voiced frames Sentence 6 Unvoiced frames Sentence 6 interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-31
SLIDE 31

31

  • 1. Introduction
  • 2. Methodology
  • 3. Databases
  • 4. Device
  • 5. Results
  • 6. Conclusion

Outline

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-32
SLIDE 32
  • 5. Conclusion
  • 1. A computational tool is presented for the recording and analysis of

speech with Parkinson’s disease

  • 2. The methodology evaluated considers

speech recorded in non- controlled noise conditions.

  • 3. It is useful the speech enhancement technique?

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-33
SLIDE 33
  • 5. Conclusion
  • 4. The incorporation to the methodology of prosody features derived from

timing, duration, and speech rate is planned for the near future.

  • 5. The use of the computer tool to follow the speech therapy of the patients,

and to asses their neurological state is also expected in the future.

interspeech-2015

jcamilo.vasquez@udea.edu.co

slide-34
SLIDE 34

Thanks!

interspeech-2015

jcamilo.vasquez@udea.edu.co