Academia.eduAcademia.edu

Outline

Title
Abstract
Introduction
Related Works A. Model Compression
Methodology
Proposed Method
Dataset and Preprocessing
Results
Future Works
Conclusions
References
All Topics
Computer Science
Artificial Intelligence

Toward Real-World Voice Disorder Classification

Profile image of Yu TsaoYu Tsao

IEEE Transactions on Biomedical Engineering

https://doi.org/10.1109/TBME.2023.3270532
visibility

…

description

12 pages

link

1 file

Abstract

Objective: Voice disorders significantly compromise individuals' ability to speak in their daily lives. Without early diagnosis and treatment, these disorders may deteriorate drastically. Thus, automatic classification systems at home are desirable for people who are inaccessible to clinical disease assessments. However, the performance of such systems may be weakened due to the constrained resources and domain mismatch between the clinical data and noisy real-world data. Methods: This study develops a compact and domain-robust voice disorder classification system to identify the utterances of health, neoplasm, and benign structural diseases. Our proposed system utilizes a feature extractor model composed of factorized convolutional neural networks and subsequently deploys domain adversarial training to reconcile the domain mismatch by extracting domain-invariant features. Results: The results show that the unweighted average recall in the noisy realworld domain improved by 13% and remained at 80% in the clinic domain with only slight degradation. The domain mismatch was effectively eliminated. Moreover, the proposed system reduced the usage of both memory and computation by over 73.9%. Conclusion: By deploying factorized convolutional neural networks and domain adversarial training, domain-invariant features can be derived for voice disorder classification with limited resources. The promising results confirm that the proposed system can significantly reduce resource consumption and improve classification accuracy by considering the domain mismatch. Significance: To the best of our knowledge, this is the first study that jointly considers real-world model compression and noise-robustness issues in voice disorder classification. The proposed system is intended for application to embedded systems with limited resources.

Related papers

Combining acoustic signals and medical records to improve pathological voice classification
Yu Tsao

APSIPA Transactions on Signal and Information Processing

This study proposes two multimodal frameworks to classify pathological voice samples by combining acoustic signals and medical records. In the first framework, acoustic signals are transformed into static supervectors via Gaussian mixture models; then, a deep neural network (DNN) combines the supervectors with the medical record and classifies the voice signals. In the second framework, both acoustic features and medical data are processed through first-stage DNNs individually; then, a second-stage DNN combines the outputs of the first-stage DNNs and performs classification. Voice samples were recorded in a specific voice clinic of a tertiary teaching hospital, including three common categories of vocal diseases, i.e. glottic neoplasm, phonotraumatic lesions, and vocal paralysis. Experimental results demonstrated that the proposed framework yields significant accuracy and unweighted average recall (UAR) improvements of 2.02–10.32% and 2.48–17.31%, respectively, compared with systems...

View PDFchevron_right
Diagnosis of Pathological Speech with Efficient and Effective Features for Long Short-Term Memory Learning
Lifong Zou

The majority of voice disorders stem from improper vocal usage. Alterations in voice quality can also serve as indicators for a broad spectrum of diseases. Particularly, the significant correlation between voice disorders and dental health underscores the need for precise diagnosis through acoustic data. This paper introduces effective and efficient features for deep learning with speech signals to distinguish between two groups: individuals with healthy voices and those with pathological voice conditions. Using a public voice database, the ten-fold test results obtained from long short-term memory networks trained on the combination of time-frequency and time-space features with a data balance strategy achieved the following metrics: accuracy = 90%, sensitivity = 93%, specificity = 87%, precision = 88%,F1score = 0.90, and area under the receiver operating characteristic curve = 0.96.

View PDFchevron_right
Voice Disorder Detection Using Long Short Term Memory (LSTM) Model
Vibhuti Gupta

2018

Automated detection of voice disorders with computational methods is a recent research area in the medical domain since it requires a rigorous endoscopy for the accurate diagnosis. Efficient screening methods are required for the diagnosis of voice disorders so as to provide timely medical facilities in minimal resources. Detecting Voice disorder using computational methods is a challenging problem since audio data is continuous due to which extracting relevant features and applying machine learning is hard and unreliable. This paper proposes a Long short term memory model (LSTM) to detect pathological voice disorders and evaluates its performance in a real 400 testing samples without any labels. Different feature extraction methods are used to provide the best set of features before applying LSTM model for classification. The paper describes the approach and experiments that show promising results with 22% sensitivity, 97% specificity and 56% unweighted average recall.

View PDFchevron_right
Leveraging Artificial Intelligence to Improve Voice Disorder Identification Through the Use of a Reliable Mobile App
Giovanna Sannino

IEEE Access

The evolution of the Internet of Things, cloud computing and wireless communication has contributed to an advance in the interconnectivity, efficiency and data accessibility in smart cities, improving environmental sustainability, quality of life and well-being, knowledge and intellectual capital. In this scenario, the satisfaction of security and privacy requirements to preserve data integrity, confidentiality and authentication is of fundamental importance. In particular, this is essential in the healthcare sector, where health-related data are considered sensitive information able to reveal confidential details about the subject. In this regard, to limit the possibility of security attacks or privacy violations, we present a reliable mobile voice disorder detection system capable of distinguishing between healthy and pathological voices by using a machine learning algorithm. This latter is totally embedded in the mobile application, so it is able to classify the voice without the necessity of transmitting user data to or storing user data on any server. A Boosted Trees algorithm was used as the classifier, opportunely trained and validated on a dataset composed of 2003 voices. The most frequently considered acoustic parameters constituted the inputs of the classifier, estimated and analyzed in real time by the mobile application.

View PDFchevron_right
Voice Disorder Identification by Using Machine Learning Techniques
Giovanna Sannino

IEEE Access

Nowadays, the use of mobile devices in the healthcare sector is increasing significantly. Mobile technologies offer not only forms of communication for multimedia content (e.g. clinical audiovisual notes and medical records) but also promising solutions for people who desire the detection, monitoring, and treatment of their health conditions anywhere and at any time. Mobile health systems can contribute to make patient care faster, better, and cheaper. Several pathological conditions can benefit from the use of mobile technologies. In this paper we focus on dysphonia, an alteration of the voice quality that affects about one person in three at least once in his/her lifetime. Voice disorders are rapidly spreading, although they are often underestimated. Mobile health systems can be an easy and fast support to voice pathology detection. The identification of an algorithm that discriminates between pathological and healthy voices with more accuracy is necessary to realize a valid and precise mobile health system. The key contribution of this paper is to investigate and compare the performance of several machine learning techniques useful for voice pathology detection. All analyses are performed on a dataset of voices selected from the Saarbruecken voice database. The results obtained are evaluated in terms of accuracy, sensitivity, specificity, and receiver operating characteristic area. They show that the best accuracy in voice diseases detection is achieved by the support vector machine algorithm or the decision tree one, depending on the features evaluated by using opportune feature selection methods. INDEX TERMS Mobile health systems, machine learning techniques, voice disorders, classification accuracy.

View PDFchevron_right
Ensemble and Multimodal Learning for Pathological Voice Classification
Yu Tsao

IEEE Sensors Letters

Voice disorders are one of the most common medical diseases in modern society, especially for those who with occupational voice demand. In this paper, we investigate a stacked ensemble learning method to classify pathological voice disorders by combining acoustic signals and medical records. In the proposed ensemble learning framework, stacked support vector machines (SVMs) form a set of weak classifiers, and a deep neural network (DNN) acts a meta leaner. Acoustic features and medical records are combined to attain better classification performance based on the high complexity of meta learner. Results showed that the proposed approach significantly outperforms individual SVM and DNN classifiers, and showed a performance improvement over the two-stage-DNN based fusion classifier. The proposed approach achieved 89.83% accuracy and 85.84% unweighted average recall in a three-disorder classification task, confirming the effectiveness of the ensemble learning for pathological voice classification.

View PDFchevron_right
Deep Learning against COVID-19: Respiratory Insufficiency Detection in Brazilian Portuguese Speech
Lucas Rafael Gris

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, 2021

Respiratory insufficiency is a symptom that requires hospitalization. This work investigates whether it is possible to detect this condition by analyzing patient's speech samples; the analysis was performed on data collected during the first wave of the COVID-19 pandemic in 2020, and thus limited to respiratory insufficiency in COVID-19 patients. For that, a dataset was created consisting of speech emissions of both COVID-19 patients affected by respiratory insufficiency and a control group. This dataset was used to build a Convolution Neural Network to detect respiratory insufficiency using speech emission MFCC representations. Methodologically, dealing with background noise was a challenge, so we also collected background noise from COVID-19 wards where patients were located. Due to the difficulty in filtering noise without eliminating crucial information, noise samples were injected in the control group data to prevent bias. Moreover, we investigated (i) two approaches to address the duration variance of audios, and (ii) the ideal number of noise samples to inject in both patients and the control group to prevent bias and overfitting. The techniques developed reached 91.66% accuracy. Thus we validated the project's Leading Hypothesis, namely that it is possible to detect respiratory insufficiency in speech utterances, under real-life environmental conditions; we believe our results justify further enquiries into the use of automated speech analysis to support health professionals in triage procedures.

View PDFchevron_right
VOICE DISORDER DETECTION USING DEEP LEARNING
IAEME Publication

IAEME PUBLICATION, 2020

The combination of artificial intelligence (AI) and the Internet of Things (IoT) has huge chances in smart health- care. Particularly during the post-COVID-19 pandemic, the need of smart healthcare has come to light more than before. A significant number of people suffer from voice disorder. This disorder can be easily healed if detected early. In this study, a voice disorder detection system within a smart healthcare framework is proposed. The inputs are obtained by the IoT, namely microphones and electroglottography (EGG) devices to capture voice and EGG signals. Spectrograms are obtained from these signals and fed into a pretrained convolutional neural Network (CNN). You may have a voice disorder if you have a problem with pitch, volume, tone, and other qualities of your voice. These problems occur when your vocal cords don’t vibrate normally. The voice is the sound that air makes when it is forced out of your lungs and passes over your vocal cords. Vocal cords are the 2 folds of tissue inside your larynx, also called the voice box. The vibration of those cords is what produces speech. Examples of voice disorders: Laryngitis: Laryngitis is when your vocal cords swell. It makes the voice sound hoarse. Or you may not be able to speak at all. Acute laryngitis happens suddenly, often because of a virus in the upper respiratory tract. Cyst: A cyst is a development that can be found under the vocal fold mucosa’s surface layer. If a void forms between the 2 vocal folds, the vocal folds cannot vibrate normally. Vocal fold Polyps: Polyps and cysts both grow from the vocal fold mucosa. Strong or fluid packed, they may expand to quite large sizes.

View PDFchevron_right
Comparative Analysis of CNN and RNN for Voice Pathology Detection
HIRA ZAHID

BioMed Research International, 2021

Diagnosis on the basis of a computerized acoustic examination may play an incredibly important role in early diagnosis and in monitoring and even improving effective pathological speech diagnostics. Various acoustic metrics test the health of the voice. The precision of these parameters also has to do with algorithms for the detection of speech noise. The idea is to detect the disease pathology from the voice. First, we apply the feature extraction on the SVD dataset. After the feature extraction, the system input goes into the 27 neuronal layer neural networks that are convolutional and recurrent neural network. We divided the dataset into training and testing, and after 10 k-fold validation, the reported accuracies of CNN and RNN are 87.11% and 86.52%, respectively. A 10-fold cross-validation is used to evaluate the performance of the classifier. On a Linux workstation with one NVidia Titan X GPU, program code was written in Python using the TensorFlow package.

View PDFchevron_right
ByoVoz Automatic Voice Condition Analysis System for the 2018 FEMH Challenge
Julian Arias-Londoño

2018 IEEE International Conference on Big Data (Big Data), 2018

This paper presents the methods and results used by the ByoVoz team for the design of an automatic voice condition analysis system, which was submitted to the 2018 Far East Memorial Hospital voice data challenge. The proposed methodology is based on a cascading scheme that firstly discriminates between pathological and normophonic voices, and then identifies the type of disorder. By using diverse feature selection techniques, a subset of complexity, spectral/cepstral and perturbation characteristics were identified for the proposed tasks. Then, several generative classification methodologies based on Gaussian Mixture Models and Gradient Boosting were employed to provide decisions about the input voices in the binary classification, and using onevs-one classification systems based on Random Forests for the categorization according to the type of disorder. By using a 4-folds cross-validation approach on the training partition a sensitivity=0.93 and specificity=0.74 were obtained. Similarly, an unweighted average recall of 0.63 and an accuracy of 66% was obtained for the identification task. Using the scoring metric proposed in the challenge the final resulting score considering both detection and identification is of 0.77.

View PDFchevron_right

References (77)

  1. J. K. Laguaite, "Adult voice screening," Journal of Speech and Hearing Disorders, vol. 37, no. 2, pp. 147-151, 1972.
  2. D. E. Morley, "A ten-year survey of speech disorders among university students," Journal of Speech and Hearing disorders, vol. 17, no. 1, pp. 25-31, 1952.
  3. L. O. Ramig and K. Verdolini, "Treatment efficacy: voice disorders," Journal of Speech, Language, and Hearing Research, vol. 41, no. 1, pp. S101-S116, 1998.
  4. N. Roy, R. M. Merrill, S. D. Gray, and E. M. Smith, "Voice disorders in the general population: prevalence, risk factors, and occupational impact," The Laryngoscope, vol. 115, no. 11, pp. 1988-1995, 2005.
  5. S. M. Cohen, "Self-reported impact of dysphonia in a primary care population: An epidemiological study," The Laryngoscope, vol. 120, no. 10, pp. 2022-2032, 2010.
  6. S. R. Best and C. Fakhry, "The prevalence, diagnosis, and management of voice disorders in a national ambulatory medical care survey (namcs) cohort," The Laryngoscope, vol. 121, no. 1, pp. 150-157, 2011.
  7. S. M. Cohen, J. Kim, N. Roy, C. Asche, and M. Courey, "Prevalence and causes of dysphonia in a large treatment-seeking population," The Laryngoscope, vol. 122, no. 2, pp. 343-348, 2012.
  8. A. A. Dibazar, S. Narayanan, and T. W. Berger, "Feature analysis for automatic detection of pathological speech," in Proceedings of the Second Joint 24th Annual Conference and the Annual Fall Meeting of the Biomedical Engineering Society][Engineering in Medicine and Biology, vol. 1. IEEE, 2002, pp. 182-183.
  9. P. Henríquez, J. B. Alonso, M. A. Ferrer, C. M. Travieso, J. I. Godino-Llorente, and F. Díaz-de María, "Characterization of healthy and pathological voice through measures based on nonlinear dynamics," IEEE transactions on audio, speech, and language processing, vol. 17, no. 6, pp. 1186-1195, 2009.
  10. G. Vaziri, F. Almasganj, and R. Behroozmand, "Pathological assessment of patients' speech signals using nonlinear dynamical analysis," Com- puters in biology and medicine, vol. 40, no. 1, pp. 54-63, 2010.
  11. J. M. Miramont, M. A. Colominas, and G. Schlotthauer, "Emulating perceptual evaluation of voice using scattering transform based features," IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 1892-1901, 2022.
  12. M. Cooke, O. Scharenborg, and B. T. Meyer, "The time course of adaptation to distorted speech," The Journal of the Acoustical Society of America, vol. 151, no. 4, pp. 2636-2646, 2022. [Online]. Available: https://doi.org/10.1121/10.0010235
  13. M. Illa, B. M. Halpern, R. van Son, L. Moro-Velázquez, and O. Scharen- borg, "Pathological voice adaptation with autoencoder-based voice con- version," arXiv preprint arXiv:2106.08427, 2021.
  14. B. M. Halpern, J. Fritsch, E. Hermann, R. van Son, O. Scharenborg, and M. Magimai-Doss, "An objective evaluation framework for pathological speech synthesis," in Speech Communication; 14th ITG Conference. VDE, 2021, pp. 1-5.
  15. J. Unger, J. Lohscheller, M. Reiter, K. Eder, C. S. Betz, and M. Schuster, "A noninvasive procedure for early-stage discrimination of malignant and precancerous vocal fold lesions based on laryngeal dynamics analysisearly-stage detection of malignant vocal fold lesions," Cancer research, vol. 75, no. 1, pp. 31-39, 2015.
  16. R. Fraile, N. Sáenz-Lechón, J. I. Godino-Llorente, V. Osma-Ruiz, and C. Fredouille, "Automatic detection of laryngeal pathologies in records of sustained vowels by means of Mel-frequency cepstral coefficient parameters and differentiation of patients by sex," Folia phoniatrica et logopaedica, vol. 61, no. 3, pp. 146-152, 2009.
  17. S. C. Costa, B. G. A. Neto, and J. M. Fechine, "Pathological voice discrimination using cepstral analysis, vector quantization and hidden markov models," in 2008 8th IEEE International Conference on BioIn- formatics and BioEngineering. IEEE, 2008, pp. 1-5.
  18. P. Harar, Z. Galaz, J. B. Alonso-Hernandez, J. Mekyska, R. Burget, and Z. Smekal, "Towards robust voice pathology detection," Neural Computing and Applications, vol. 32, no. 20, pp. 15 747-15 757, 2020.
  19. K. Umapathy, S. Krishnan, V. Parsa, and D. G. Jamieson, "Discrimi- nation of pathological voices using a time-frequency approach," IEEE Transactions on Biomedical Engineering, vol. 52, no. 3, pp. 421-430, 2005.
  20. M. Pützer and W. Wokurek, "Electroglottographic and acoustic parametrization of phonatory quality provide voice profiles of patho- logical speakers," Journal of Voice, 2021.
  21. C. Zhou, Y. Wu, Z. Fan, X. Zhang, D. Wu, and Z. Tao, "Gammatone spectral latitude features extraction for pathological voice detection and classification," Applied Acoustics, vol. 185, p. 108417, 2022.
  22. J. I. Godino-Llorente, P. Gomez-Vilda, and M. Blanco-Velasco, "Dimen- sionality reduction of a pathological voice quality assessment system based on gaussian mixture models and short-term cepstral parameters," IEEE transactions on biomedical engineering, vol. 53, no. 10, pp. 1943- 1953, 2006.
  23. M. K. Arjmandi and M. Pooyan, "An optimum algorithm in pathological voice quality assessment using wavelet-packet-based features, linear discriminant analysis and support vector machine," Biomedical Signal Processing and Control, vol. 7, no. 1, pp. 3-19, 2012.
  24. M. Markaki and Y. Stylianou, "Voice pathology detection and discrim- ination based on modulation spectral features," IEEE Transactions on audio, speech, and language processing, vol. 19, no. 7, pp. 1938-1948, 2011.
  25. I. Hammami, L. Salhi, and S. Labidi, "Pathological voices detection using support vector machine," in 2016 2nd International Conference on Advanced Technologies for Signal and Image Processing (ATSIP). IEEE, 2016, pp. 662-666.
  26. L. Verde, G. De Pietro, and G. Sannino, "Voice disorder identification by using machine learning techniques," IEEE access, vol. 6, pp. 16 246- 16 255, 2018.
  27. M. Pishgar, F. Karim, S. Majumdar, and H. Darabi, "Pathological voice classification using Mel-cepstrum vectors and support vector machine," arXiv preprint arXiv:1812.07729, 2018.
  28. J. D. Arias-Londoño, J. I. Godino-Llorente, M. Markaki, and Y. Stylianou, "On combining information from modulation spectra and Mel-frequency cepstral coefficients for automatic detection of patholog- ical voices," Logopedics Phoniatrics Vocology, vol. 36, no. 2, pp. 60-69, 2011.
  29. J. D. Arias-Londoño, J. I. Godino-Llorente, N. Sáenz-Lechón, V. Osma- Ruiz, and G. Castellanos-Domínguez, "Automatic detection of patho- logical voices using complexity measures, noise parameters, and Mel- cepstral coefficients," IEEE Transactions on biomedical engineering, vol. 58, no. 2, pp. 370-379, 2010.
  30. M. Dahmani and M. Guerti, "Recurrence quantification analysis of glottal signal as non linear tool for pathological voice assessment and classification." Int. Arab J. Inf. Technol., vol. 17, no. 6, pp. 857-866, 2020.
  31. A. Basalamah, M. Hasan, S. Bhowmik, and S. A. Shahriyar, "A highly accurate dysphonia detection system using linear discriminant analysis," COMPUTER SYSTEMS SCIENCE AND ENGINEERING, vol. 44, no. 3, pp. 1921-1938, 2023.
  32. H. Wu, J. Soraghan, A. Lowit, and G. Di Caterina, "Convolutional neural networks for pathological voice detection," in 2018 40th annual international conference of the ieee engineering in medicine and biology society (EMBC). IEEE, 2018, pp. 1-4.
  33. V. Gupta, "Voice disorder detection using long short term memory (LSTM) model," arXiv preprint arXiv:1812.01779, 2018.
  34. S.-H. Fang, Y. Tsao, M.-J. Hsiao, J.-Y. Chen, Y.-H. Lai, F.-C. Lin, and C.-T. Wang, "Detection of pathological voice using cepstrum vectors: A deep learning approach," Journal of Voice, vol. 33, no. 5, pp. 634-641, 2019.
  35. C.-H. Hung, S.-S. Wang, C.-T. Wang, and S.-H. Fang, "Using sincnet for learning pathological voice disorders," Sensors, vol. 22, no. 17, p. 6634, 2022.
  36. J.-Y. Lee, "Experimental evaluation of deep learning methods for an intelligent pathological voice detection system using the saarbruecken voice database," Applied Sciences, vol. 11, no. 15, p. 7149, 2021.
  37. W. Ariyanti, T. Hussain, J.-C. Wang, C.-T. Wang, S.-H. Fang, and Y. Tsao, "Ensemble and multimodal learning for pathological voice classification," IEEE Sensors Letters, vol. 5, no. 7, pp. 1-4, 2021.
  38. K. G. Dávid Sztahó and T. M. Gábriel, "Deep learning solution for pathological voice detection using lstm-based autoencoder hybrid with multi-task learning," 2021.
  39. T. Lee, Y. Liu, P.-W. Huang, J.-T. Chien, W. K. Lam, Y. T. Yeung, T. K. Law, K. Y. Lee, A. P.-H. Kong, and S.-P. Law, "Automatic speech recognition for acoustical analysis and assessment of cantonese pathological voice and speech," in 2016 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2016, pp. 6475-6479.
  40. Y. Liu, T. Lee, T. Law, and K. Y.-S. Lee, "Acoustical assessment of voice disorder with continuous speech using asr posterior features," IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 27, no. 6, pp. 1047-1059, 2019.
  41. S.-H. Fang, C.-T. Wang, J.-Y. Chen, Y. Tsao, and F.-C. Lin, "Combining acoustic signals and medical records to improve pathological voice clas- sification," APSIPA Transactions on Signal and Information Processing, vol. 8, 2019.
  42. T. Kojima, S. Fujimura, K. Hasebe, Y. Okanoue, O. Shuya, R. Yuki, K. Shoji, R. Hori, Y. Kishimoto, and K. Omori, "Objective assessment of pathological voice using artificial intelligence based on the grbas scale," Journal of Voice, 2021.
  43. Y.-T. Hsu, Z. Zhu, C.-T. Wang, S.-H. Fang, F. Rudzicz, and Y. Tsao, "Robustness against the channel effect in pathological voice detection," arXiv preprint arXiv:1811.10376, 2018.
  44. Z. Fan, Y. Wu, C. Zhou, X. Zhang, and Z. Tao, "Class-imbalanced voice pathology detection and classification using fuzzy cluster oversampling method," Applied Sciences, vol. 11, no. 8, p. 3450, 2021.
  45. Q. Jinyang, Z. Denghuang, F. Ziqi, W. Di, X. Yishen, and T. Zhi, "Patho- logical voice feature generation using generative adversarial network," in 2021 International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence (ICSMD). IEEE, 2021, pp. 1-6.
  46. G. Muhammad, S. M. M. Rahman, A. Alelaiwi, and A. Alamri, "Smart health solution integrating iot and cloud: A case study of voice pathology monitoring," IEEE Communications Magazine, vol. 55, no. 1, pp. 69-73, 2017.
  47. G. Muhammad, M. F. Alhamid, M. Alsulaiman, and B. Gupta, "Edge computing with cloud for voice disorder assessment and treatment," IEEE Communications Magazine, vol. 56, no. 4, pp. 60-65, 2018.
  48. M. S. Hossain, G. Muhammad, and A. Alamri, "Smart healthcare monitoring: a voice pathology detection paradigm for smart cities," Multimedia Systems, vol. 25, pp. 565-575, 2019.
  49. S. U. Amin, M. S. Hossain, G. Muhammad, M. Alhussein, and M. A. Rahman, "Cognitive smart healthcare for pathology detection and mon- itoring," IEEE Access, vol. 7, pp. 10 745-10 753, 2019.
  50. A. Ramalingam, S. Kedari, and C. Vuppalapati, "Ieee femh voice data challenge 2018," in 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018, pp. 5272-5276.
  51. M. Pham, J. Lin, and Y. Zhang, "Diagnosing voice disorder with machine learning," in 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018, pp. 5263-5266.
  52. T. Grzywalski, A. Maciaszek, A. Biniakowski, J. Orwat, S. Drgas, M. Piecuch, R. Belluzzo, K. Joachimiak, D. Niemiec, J. Ptaszynski et al., "Parameterization of sequence of mfccs for dnn-based voice disorder detection," in 2018 IEEE International conference on big data (big data). IEEE, 2018, pp. 5247-5251.
  53. C. Bhat and S. K. Kopparapu, "FEMH voice data challenge: Voice disorder detection and classification using acoustic descriptors," in 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018, pp. 5233-5237.
  54. K. Degila, R. Errattahi, and A. El Hannani, "The UCD system for the 2018 FEMH voice data challenge," in 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018, pp. 5242-5246.
  55. J. D. Arias-Londoño, J. A. Gómez-García, L. Moro-Velázquez, and J. I. Godino-Llorente, "Byovoz automatic voice condition analysis system for the 2018 FEMH challenge," in 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018, pp. 5228-5232.
  56. K. A. Islam, D. Perez, and J. Li, "A transfer learning approach for the 2018 FEMH voice data challenge," in 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018, pp. 5252-5257.
  57. B. Jacob, S. Kligys, B. Chen, M. Zhu, M. Tang, A. Howard, H. Adam, and D. Kalenichenko, "Quantization and training of neural networks for efficient integer-arithmetic-only inference," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 2704- 2713.
  58. Y.-C. Lin, C. Yu, Y.-T. Hsu, S.-W. Fu, Y. Tsao, and T.-W. Kuo, "Seofp-net: Compression and acceleration of deep neural networks for speech enhancement using sign-exponent-only floating-points," IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 1016-1031, 2021.
  59. G. Hinton, O. Vinyals, and J. Dean, "Distilling the knowledge in a neural network," arXiv preprint arXiv:1503.02531, 2015.
  60. M. Wang, B. Liu, and H. Foroosh, "Factorized convolutional neural networks," in Proceedings of the IEEE International Conference on Computer Vision Workshops, 2017, pp. 545-553.
  61. E. Romera, J. M. Alvarez, L. M. Bergasa, and R. Arroyo, "Erfnet: Effi- cient residual factorized convnet for real-time semantic segmentation," IEEE Transactions on Intelligent Transportation Systems, vol. 19, no. 1, pp. 263-272, 2017.
  62. A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, "Mobilenets: Efficient convo- lutional neural networks for mobile vision applications," arXiv preprint arXiv:1704.04861, 2017.
  63. K. Bousmalis, N. Silberman, D. Dohan, D. Erhan, and D. Krishnan, "Unsupervised pixel-level domain adaptation with generative adversarial networks," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 3722-3731.
  64. E. Tzeng, J. Hoffman, N. Zhang, K. Saenko, and T. Darrell, "Deep domain confusion: Maximizing for domain invariance," arXiv preprint arXiv:1412.3474, 2014.
  65. H.-Y. Lin, H.-H. Tseng, X. Lu, and Y. Tsao, "Unsupervised noise adap- tive speech enhancement by discriminator-constrained optimal trans- port," Advances in Neural Information Processing Systems, vol. 34, 2021.
  66. H. Hu, S. M. Siniscalchi, C.-H. H. Yang, and C.-H. Lee, "A variational bayesian approach to learning latent variables for acoustic knowledge transfer," arXiv preprint arXiv:2110.08598, 2021.
  67. Y. Ganin, E. Ustinova, H. Ajakan, P. Germain, H. Larochelle, F. Lavi- olette, M. Marchand, and V. Lempitsky, "Domain-adversarial training of neural networks," The journal of machine learning research, vol. 17, no. 1, pp. 2096-2030, 2016.
  68. P. O. Pinheiro, "Unsupervised domain adaptation with similarity learn- ing," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 8004-8013.
  69. M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, "Mobilenetv2: Inverted residuals and linear bottlenecks," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4510-4520.
  70. A. Howard, M. Sandler, G. Chu, L.-C. Chen, B. Chen, M. Tan, W. Wang, Y. Zhu, R. Pang, V. Vasudevan et al., "Searching for mobilenetv3," in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 1314-1324.
  71. I. Redko, E. Morvant, A. Habrard, M. Sebban, and Y. Bennani, "A survey on domain adaptation theory: learning bounds and theoretical guarantees," arXiv preprint arXiv:2004.11829, 2020.
  72. Y. Zhang, J. Qian, X. Zhang, Y. Xu, and Z. Tao, "Pathological voice detection using joint subsapce transfer learning," Applied Sciences, vol. 12, no. 16, p. 8129, 2022.
  73. S. Ruder, "An overview of gradient descent optimization algorithms," arXiv preprint arXiv:1609.04747, 2016.
  74. S. Ioffe and C. Szegedy, "Batch normalization: Accelerating deep network training by reducing internal covariate shift," in International conference on machine learning. PMLR, 2015, pp. 448-456.
  75. B. Xu, N. Wang, T. Chen, and M. Li, "Empirical evaluation of rectified activations in convolutional network," arXiv preprint arXiv:1505.00853, 2015.
  76. D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," arXiv preprint arXiv:1412.6980, 2014.
  77. T.-T. Wong, "Performance evaluation of classification algorithms by k- fold and leave-one-out cross validation," Pattern Recognition, vol. 48, no. 9, pp. 2839-2846, 2015.

Related papers

Automatic Voice Disorder Detection Using Self-Supervised Representations
David Vergara Martinez

IEEE Access

Many speech features and models, including Deep Neural Networks (DNN), are used for classification tasks between healthy and pathological speech with the Saarbruecken Voice Database (SVD). However, accuracy values of 80.71% for phrases or 82.8% for vowels /aiu/ are the highest reported for audio samples in SVD when the evaluation includes the wide amount of pathologies in the database, instead of a selection of some pathologies. This paper targets this top performance in the state-of-the-art Automatic Voice Disorder Detection (AVDD) systems. In the framework of a DNN-based AVDD system we study the capability of Self-Supervised (SS) representation learning for describing discriminative cues between healthy and pathological speech. The system processes the SS temporal sequence of features with a single feed-forward layer and Class-Token (CT) Transformer for obtaining the classification between healthy and pathological speech. Furthermore, there is evaluated a suitable data extension of the training set with out-ofdomain data is also evaluated to deal with the low availability of data for using DNN-based models in voice pathology detection. Experimental results using audio samples corresponding to phrases in the SVD dataset, including all pathologies available, show classification accuracy values until 93.36%. This means that the proposed AVDD system achieved accuracy improvements of 4.1% without the training data extension, and 15.62% after the training data extension compared to the baseline system. Beyond the novelty of using SS representations for AVDD, the fact of obtaining accuracies over 90% in these conditions and using the whole set of pathologies in the SVD is a milestone for voice disorder-related research. Furthermore, the study on the amount of in-domain data in the training set related to the system performance show guidance for the data preparation stage. Lessons learned in this work suggest guidelines for taking advantage of DNN, to boost the performance in developing automatic systems for diagnosis, treatment, and monitoring of voice pathologies. INDEX TERMS Voice disorder, pathological speech, Saarbruecken voice database, advanced voice function assessment database, self-supervised, class token, transformer, deep neural networks.

View PDFchevron_right
Transfer Learning with AudioSet to Voice Pathologies Identification in Continuous Speech
victor guedes

Procedia Computer Science, 2019

The classification of pathological diseases with the implementation of concepts of Deep Learning has been increasing considerably in recent times. Among the works developed there are good results for the classification in sustained speech with vowels, but few related works for the classification in continuous speech. This work uses the German Saarbrücken Voice Database with the phrase "Guten Morgen, wie geht es Ihnen?" to classify four classes: dysphonia, laryngitis, paralysis of vocal cords and healthy voices. Transfer learning concepts were used with the AudioSet database. Two models were developed based on Long-Short-Term-Memory and Convolutional Network for classification of extracted embeddings and comparison of the best results, using crossvalidation. The final results allowed to obtaining 40% of f1-score for the four classes, 66% f1-score for Dysphonia x Healthy, 67% for Laryngitis x healthy and 80% for Paralysis x Healthy.

View PDFchevron_right
Utilizing Wav2Vec In Database-Independent Voice Disorder Detection
Sudarsana Kadiri

ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Automatic detection of voice disorders from acoustic speech signals can help to improve reliability of medical diagnosis. However, the real-life environment in which speech signals are recorded for diagnosis can be different from the environment in which the detection system's training data was originally collected. This mismatch between the recording conditions can decrease detection performance in practical scenarios. In this work, we propose to use a pre-trained wav2vec 2.0 model as a feature extractor to build automatic detection systems for voice disorders. The embeddings from the first layers of the context network contain information about phones, and these features are useful in voice disorder detection. We evaluate the performance of the wav2vec features in single-database and crossdatabase scenarios to study their generalizability to unseen speakers and recording conditions. The results indicate that the wav2vec features generalize better than popular spectral and cepstral baseline features.

View PDFchevron_right
Robustness against the channel effect in pathological voice detection
Yu Tsao

ArXiv, 2018

Many people are suffering from voice disorders, which can adversely affect the quality of their lives. In response, some researchers have proposed algorithms for automatic assessment of these disorders, based on voice signals. However, these signals can be sensitive to the recording devices. Indeed, the channel effect is a pervasive problem in machine learning for healthcare. In this study, we propose a detection system for pathological voice, which is robust against the channel effect. This system is based on a bidirectional LSTM network. To increase the performance robustness against channel mismatch, we integrate domain adversarial training (DAT) to eliminate the differences between the devices. When we train on data recorded on a high-quality microphone and evaluate on smartphone data without labels, our robust detection system increases the PR-AUC from 0.8448 to 0.9455 (and 0.9522 with target sample labels). To the best of our knowledge, this is the first study applying unsuper...

View PDFchevron_right
Acoustic Feature Extraction with Interpretable Deep Neural Network for Neurodegenerative Related Disorder Classification
Bahman Mirheidari

Interspeech 2020, 2020

Speech-based automatic approaches for detecting neurodegenerative disorders (ND) and mild cognitive impairment (MCI) have received more attention recently due to being non-invasive and potentially more sensitive than current pen-and-paper tests. The performance of such systems is highly dependent on the choice of features in the classification pipeline. In particular for acoustic features, arriving at a consensus for a best feature set has proven challenging. This paper explores using deep neural network for extracting features directly from the speech signal as a solution to this. Compared with hand-crafted features, more information is present in the raw waveform, but the feature extraction process becomes more complex and less interpretable which is often undesirable in medical domains. Using a Sinc-Net as a first layer allows for some analysis of learned features. We propose and evaluate the Sinc-CLA (with SincNet, Convolutional, Long Short-Term Memory and Attention layers) as a task-driven acoustic feature extractor for classifying MCI, ND and healthy controls (HC). Experiments are carried out on an inhouse dataset. Compared with the popular hand-crafted feature sets, the learned task-driven features achieve a superior classification accuracy. The filters of the SincNet is inspected and acoustic differences between HC, MCI and ND are found.

View PDFchevron_right

Related topics

  • Computer Science
  • Artificial Intelligence
  • Biomedical Engineering
  • Robustness (evolution)
  • Electrical and Electronic Engine...
    • 580 California St., Suite 400
    San Francisco, CA, 94104
    © 2025 Academia. All rights reserved