Towards robust voice pathology detection

Investigation of supervised deep learning, gradient boosting, and anomaly detection approaches across four databases
  • Pavol HararEmail author
  • Zoltan Galaz
  • Jesus B. Alonso-Hernandez
  • Jiri Mekyska
  • Radim Burget
  • Zdenek Smekal
S.I.: Advances in Bio-Inspired Intelligent Systems


Automatic objective non-invasive detection of pathological voice based on computerized analysis of acoustic signals can play an important role in early diagnosis, progression tracking, and even effective treatment of pathological voices. In search towards such a robust voice pathology detection system, we investigated three distinct classifiers within supervised learning and anomaly detection paradigms. We conducted a set of experiments using a variety of input data such as raw waveforms, spectrograms, mel-frequency cepstral coefficients (MFCC), and conventional acoustic (dysphonic) features (AF). In comparison with previously published works, this article is the first to utilize combination of four different databases comprising normophonic and pathological recordings of sustained phonation of the vowel /a/ unrestricted to a subset of vocal pathologies. Furthermore, to our best knowledge, this article is the first to explore gradient-boosted trees and deep learning for this application. The following best classification performances measured by F1 score on dedicated test set were achieved: XGBoost (0.733) using AF and MFCC, DenseNet (0.621) using MFCC, and Isolation Forest (0.610) using AF. Even though these results are of exploratory character, conducted experiments do show promising potential of gradient boosting and deep learning methods to robustly detect voice pathologies.


Voice pathology detection Deep learning Gradient boosting Anomaly detection 



This study was funded by the grant of the Czech Ministry of Health 16-30805A (Effects of non-invasive brain stimulation on hypokinetic dysarthria, micrographia, and brain plasticity in patients with Parkinson’s disease) and the following projects: SIX (CZ.1.05/2.1.00/03.0072) and LO1401. For the research, infrastructure of the SIX Center was used. The authors (P. Harar, Z. Galaz) of this study also acknowledge the financial support of Erwin Schrödinger International Institute for Mathematics and Physics during their stay at the “Systematic approaches to deep learning methods for audio” workshop held from 11 September, 2017, to 15 September, 2017, in Vienna.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.


  1. 1.
    Ali Z, Alsulaiman M, Muhammad G, Elamvazuthi I, Al-nasheri A, Mesallam TA, Farahat M, Malki KH (2017) Intra-and inter-database study for arabic, english, and german databases: do conventional speech features detect voice pathology? J Voice 31(3):e381–e386CrossRefGoogle Scholar
  2. 2.
    Ali Z, Muhammad G, Alhamid MF (2017) An automatic health monitoring system for patients suffering from voice complications in smart cities. IEEE Access 5:3900–3908CrossRefGoogle Scholar
  3. 3.
    Al-nasheri A, Muhammad G, Alsulaiman M, Ali Z (2017) Investigation of voice pathology detection and classification on different frequency regions using correlation functions. J Voice 31(1):3–15CrossRefGoogle Scholar
  4. 4.
    Al-nasheri A, Muhammad G, Alsulaiman M, Ali Z, Malki K, Mesallam T, Farahat M (2017) Voice pathology detection and classification using auto-correlation and entropy features in different frequency regions. IEEE Access PP(99):1–1. Google Scholar
  5. 5.
    Al-nasheri A, Muhammad G, Alsulaiman M, Ali Z, Mesallam TA, Farahat M, Malki KH, Bencherif MA (2017) An investigation of multidimensional voice program parameters in three different databases for voice pathology detection and classification. J Voice 31(1):113–e119CrossRefGoogle Scholar
  6. 6.
    Al-nasheri A, Ali Z, Muhammad G, Alsulaiman M (2014) Voice pathology detection using auto-correlation of different filters bank. In: 2014 IEEE/ACS 11th international conference on computer systems and applications (AICCSA), pp 50–55. IEEEGoogle Scholar
  7. 7.
    Amami R, Smiti A (2017) An incremental method combining density clustering and support vector machines for voice pathology detection. Comput Electr Eng 57:257–265CrossRefGoogle Scholar
  8. 8.
    Arias-Londoño JD, Godino-Llorente JI, Markaki M, Stylianou Y (2011) On combining information from modulation spectra and mel-frequency cepstral coefficients for automatic detection of pathological voices. Logop Phoniatr Vocol 36(2):60–69CrossRefGoogle Scholar
  9. 9.
    Armstrong D, Gosling A, Weinman J, Marteau T (1997) The place of inter-rater reliability in qualitative research: an empirical study. Sociology 31(3):597–606CrossRefGoogle Scholar
  10. 10.
    Brabenec L, Mekyska J, Galaz Z, Rektorova I (2017) Speech disorders in parkinsons disease: early diagnostics and effects of medication and brain stimulation. J Neural Transm 124(3):303–334CrossRefGoogle Scholar
  11. 11.
    Breiman L (2001) Random forests. Mach Learn 45(1):5–32CrossRefzbMATHGoogle Scholar
  12. 12.
    Chen T, Guestrin C (2016) Xgboost: a scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pp 785–794. ACMGoogle Scholar
  13. 13.
    Chollet F et al (2015) Keras: Deep learning library for theano and tensorflow.
  14. 14.
    Dahmani M, Guerti M (2017) Vocal folds pathologies classification using naïve bayes networks. In: 2017 6th international conference on systems and control (ICSC), pp 426–432. IEEEGoogle Scholar
  15. 15.
    De Bodt MS, Wuyts FL, Van de Heyning PH, Croux C (1997) Test–retest study of the grbas scale: influence of experience and professional background on perceptual rating of voice quality. J Voice 11(1):74–80CrossRefGoogle Scholar
  16. 16.
    Dejonckere PH, Bradley P, Clemente P, Cornut G, Crevier-Buchman L, Friedrich G, Van De Heyning P, Remacle M, Woisard V (2001) A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Eur Arch Otorhinolaryngol 258(2):77–82CrossRefGoogle Scholar
  17. 17.
    Eskidere Ö, Gürhanlı A (2015) Voice disorder classification based on multitaper mel frequency cepstral coefficients features. Comput Math Methods Med. Google Scholar
  18. 18.
    Eye M, Infirmary E (1994) Voice disorders database, version. 1.03 (cd-rom). Kay Elemetrics Corporation, Lincoln ParkGoogle Scholar
  19. 19.
    Gerratt BR, Kreiman J, Antonanzas-Barroso N, Berke GS (1993) Comparing internal and external standards in voice quality judgments. J Speech Hear Res 36(1):14–20CrossRefGoogle Scholar
  20. 20.
    Godino-Llorente JI, Gómez-Vilda P, Cruz-Roldán F, Blanco-Velasco M, Fraile R (2010) Pathological likelihood index as a measurement of the degree of voice normality and perceived hoarseness. J Voice 24(6):667–677CrossRefGoogle Scholar
  21. 21.
    Gwet KL (2014) Handbook of inter-rater reliability: the definitive guide to measuring the extent of agreement among raters. Advanced Analytics LLC, MontgomeryGoogle Scholar
  22. 22.
    Harar P, Alonso-Hernandezy JB, Mekyska J, Galaz Z, Burget R, Smekal Z (2017) Voice pathology detection using deep learning: a preliminary study. In: 2017 international conference and workshop on bioinspired intelligence (IWOBI), pp 1–4. IEEEGoogle Scholar
  23. 23.
    Hartigan JA, Wong MA (1979) Algorithm as 136: a k-means clustering algorithm. J Roy Stat Soc Ser C (Appl Stat) 28(1):100–108zbMATHGoogle Scholar
  24. 24.
    Hearst MA, Dumais ST, Osuna E, Platt J, Scholkopf B (1998) Support vector machines. IEEE Intell Syst Their Appl 13(4):18–28CrossRefGoogle Scholar
  25. 25.
    Hemmerling D (2017) Voice pathology distinction using autoassociative neural networks. In: 2017 25th European signal processing conference (EUSIPCO), pp 1844–1847. IEEEGoogle Scholar
  26. 26.
    Hemmerling D, Skalski A, Gajda J (2016) Voice data mining for laryngeal pathology assessment. Comput Biol Med 69:270–276CrossRefGoogle Scholar
  27. 27.
    Hillenbrand J, Houde RA (1996) Acoustic correlates of breathy vocal quality: dysphonic voices and continuous speech. J Speech Hear Res 39(2):311–321CrossRefGoogle Scholar
  28. 28.
    Hossain MS, Muhammad G (2016) Healthcare big data voice pathology assessment framework. IEEE Access 4:7806–7815CrossRefGoogle Scholar
  29. 29.
    Huang GB, Zhu QY, Siew CK (2006) Extreme learning machine: theory and applications. Neurocomputing 70(1):489–501CrossRefGoogle Scholar
  30. 30.
    Huang G, Liu Z, Weinberger KQ, van der Maaten L (2016) Densely connected convolutional networks. arXiv preprint arXiv:1608.06993
  31. 31.
    Kingma D, Ba J (2014) Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980
  32. 32.
    Kreiman J, Gerratt BR, Kempster GB, Erman A, Berke GS (1993) Perceptual evaluation of voice quality: review, tutorial, and a framework for future research. J Speech Hear Res 36(1):21–40CrossRefGoogle Scholar
  33. 33.
    Little M, McSharry P, Hunter E, Spielman J, Ramig L (2009) Suitability of dysphonia measurements for telemonitoring of Parkinson’s disease. IEEE T Bio-Med Eng 56(4):1015–1022CrossRefGoogle Scholar
  34. 34.
    Liu FT, Ting KM, Zhou ZH (2008) Isolation forest. In: Eighth IEEE international conference on data mining, 2008. ICDM’08, pp 413–422. IEEEGoogle Scholar
  35. 35.
    Liu FT, Ting KM, Zhou ZH (2012) Isolation-based anomaly detection. ACM Trans Knowl Discov Data (TKDD) 6(1):3Google Scholar
  36. 36.
    Martínez D, Lleida E, Ortega A, Miguel A, Villalba J (2012) Voice pathology detection on the saarbrücken voice database with calibration and fusion of scores using multifocal toolkit. In: Advances in speech and language technologies for Iberian Languages, pp 99–109. SpringerGoogle Scholar
  37. 37.
    Mehta DD, Hillman RE (2008) Voice assessment: updates on perceptual, acoustic, aerodynamic, and endoscopic imaging methods. Curr Opin Otolaryngol Head Neck Surg 16(3):211CrossRefGoogle Scholar
  38. 38.
    Mekyska J, Janousova E, Gomez-Vilda P, Smekal Z, Rektorova I, Eliasova I, Kostalova M, Mrackova M, Alonso-Hernandez JB, Faundez-Zanuy M et al (2015) Robust and complex approach of pathological speech signal analysis. Neurocomputing 167:94–111CrossRefGoogle Scholar
  39. 39.
    Mekyska J, Galaz Z, Mzourek Z, Smekal Z, Rektorova I (2015) Assessing progress of Parkinson’s using acoustic analysis of phonation. In: 2015 International work conference on bioinspired intelligence (IWOBI), pp 115–122.
  40. 40.
    Mekyska J, Smekal Z, Galaz Z, Mzourek Z, Rektorova I, Faundez-Zanuy M, López-de Ipiña K (2016) Recent advances in nonlinear speech processing, chap. Perceptual features as markers of Parkinson’s disease: the issue of clinical interpretability, pp 83–91. Springer, Cham.
  41. 41.
    Mesallam TA, Farahat M, Malki KH, Alsulaiman M, Ali Z, Al-nasheri A, Muhammad G (2017) Development of the Arabic voice pathology database and its evaluation by using speech features and machine learning algorithms. J Healthc Eng. Google Scholar
  42. 42.
    Michaelis D, Gramss T, Strube HW (1997) Glottal-to-noise excitation ratio-a new measure for describing pathological voices. Acta Acust United Acust 83(4):700–706Google Scholar
  43. 43.
    Muhammad G, Alhamid MF, Hossain MS, Almogren AS, Vasilakos AV (2017) Enhanced living by assessing voice pathology using a co-occurrence matrix. Sensors 17(2):267CrossRefGoogle Scholar
  44. 44.
    Muhammad G, Alsulaiman M, Ali Z, Mesallam TA, Farahat M, Malki KH, Al-nasheri A, Bencherif MA (2017) Voice pathology detection using interlaced derivative pattern on glottal source excitation. Biomed Signal Process Control 31:156–164CrossRefGoogle Scholar
  45. 45.
    Murphy KP (2006) Naive bayes classifiers. University of British ColumbiaGoogle Scholar
  46. 46.
    Oates J (2009) Auditory-perceptual evaluation of disordered voice quality. Folia Phoniatr Logop 61(1):49–56CrossRefGoogle Scholar
  47. 47.
    Parsa V, Jamieson DG (2003) Identification of pathological voices using glottal noise measures. J Speech Lang Hear Res 23(2):469–485CrossRefGoogle Scholar
  48. 48.
    Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V, Vanderplas J, Passos A, Cournapeau D, Brucher M, Perrot M, Duchesnay E (2011) Scikit-learn: machine learning in python. J Mach Learn Res 12:2825–2830MathSciNetzbMATHGoogle Scholar
  49. 49.
    Pimentel MA, Clifton DA, Clifton L, Tarassenko L (2014) A review of novelty detection. Sig Process 99:215–249CrossRefGoogle Scholar
  50. 50.
    Reynolds D (2015) Gaussian mixture models. Encyclopedia of biometrics, pp 827–832Google Scholar
  51. 51.
    Sabir B, Rouda F, Khazri Y, Touri B, Moussetad M (2017) Improved algorithm for pathological and normal voices identification. Int J Electr Comput Eng (IJECE) 7(1):238–243CrossRefGoogle Scholar
  52. 52.
    Saldanha JC, Ananthakrishna T, Pinto R (2014) Vocal fold pathology assessment using mel-frequency cepstral coefficients and linear predictive cepstral coefficients features. J Med Imaging Health Inf 4(2):168–173CrossRefGoogle Scholar
  53. 53.
    Schalkoff RJ (1997) Artificial neural networks, vol 1. McGraw-Hill, New YorkzbMATHGoogle Scholar
  54. 54.
    Song P (2013) Assessment of vocal cord function and voice disorders. In: Principles and practice of interventional pulmonology, pp 137–149. SpringerGoogle Scholar
  55. 55.
    Souissi N, Cherif A (2015) Dimensionality reduction for voice disorders identification system based on mel frequency cepstral coefficients and support vector machine. In: 2015 7th international conference on modelling, identification and control (ICMIC), pp 1–6. IEEEGoogle Scholar
  56. 56.
    Souissi N, Cherif A (2016) Speech recognition system based on short-term cepstral parameters, feature reduction method and artificial neural networks. In: 2016 2nd international conference on advanced technologies for signal and image processing (ATSIP), pp 667–671. IEEEGoogle Scholar
  57. 57.
    Stathopoulos ET, Huber JE, Sussman JE (2011) Changes in acoustic characteristics of the voice across the life span: measures from individuals 4–93 years of age. J Speech Lang Hear Res 54(4):1011–1021CrossRefGoogle Scholar
  58. 58.
    Teager H (1980) Some observations on oral air flow during phonation. IEEE Trans Acoust Speech Signal Process 28(5):599–601CrossRefGoogle Scholar
  59. 59.
    Titze IR (1994) Principles of voice production. Prentice-Hall, Englewood CliffsGoogle Scholar
  60. 60.
    Tsanas A, Little MA, McSharry PE, Ramig LO (2010) Nonlinear speech analysis algorithms mapped to a standard metric achieve clinically useful quantification of average Parkinson’s disease symptom severity. J R Soc Interface 8(59):842–855CrossRefGoogle Scholar
  61. 61.
    Uloza V, Vegiene A, Saferis V (2015) Correlation between the quantitative video laryngostroboscopic measurements and parameters of multidimensional voice assessment. Biomed Signal Process Control 17(Suppl C):3–10CrossRefGoogle Scholar
  62. 62.
    Woldert-Jokisz B (2007) Saarbruecken voice databaseGoogle Scholar
  63. 63.
    Wyse L (2017) Audio spectrogram representations for processing with convolutional neural networks. arXiv preprint arXiv:1706.09559

Copyright information

© The Natural Computing Applications Forum 2018

Authors and Affiliations

  1. 1.Brno University of TechnologyBrnoCzech Republic
  2. 2.Institute for Technological Development and Innovation in Communications (IDeTIC)University of Las Palmas de Gran CanariaLas Palmas de Gran CanariaSpain

Personalised recommendations