Methodological issues in the development of automatic systems for voice pathology detection

Sáenz-Lechón, Nicolás; Godino-Llorente, Juan Ignacio; Osma-Ruiz, Víctor; Gómez-Vilda, Pedro

doi:10.1016/j.bspc.2006.06.003

Cited by 140 publications

(87 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The considered statistics are mean value, standard deviation, skewness and kurtosis, thus each recording will be represented by a total of 40 features (four statistics on ten features). The validation of the system's performance is made by the division of the data into 70% for training and 30% for testing, following the methodology exposed in [17]. The 70% of the data are used for the feature selection and for training the classifier and the remaining 30% of the data are used for testing; the different subsets for training and testing are randomly formed ten times.…”

Section: Methodsmentioning

confidence: 99%

“…The features selection process is applied again and the decision about whether a speech recording is from PPD or CS is taken with a SVM. The results are presented according to [17], indicating accuracy rates, specificity and sensitivity. Specificity indicates the probability of a healthy register to be correctly detected and sensitivity is the probability of a pathological signal to be correctly classified.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Analysis of Speech from People with Parkinson’s Disease through Nonlinear Dynamics

Orozco-Arroyave

Arias-Londoño

Vargas-Bonilla

et al. 2013

Advances in Nonlinear Speech Processing

View full text Add to dashboard Cite

Abstract. Different characterization approaches, including nonlinear dynamics (NLD), have been addressed for the automatic detection of PD; however, the obtained discrimination capability when only NLD features are considered has not been evaluated yet.This paper evaluates the discrimination capability of a set with ten different NLD features in the task of automatic classification of speech signals from people with Parkinson's disease (PPD) and a control set (CS). The experiments presented in this paper are performed considering the five Spanish vowels uttered by 20 PPD and 20 people from the CS.According the results, it is possible to achieve accuracy rates of up to 76,81% considering only utterances from the vowel /i/. When features calculated from the five Spanish vowels are combined, the performance of the system is not improved, indicating that the inclusion of more NLD features to the system does not guarantee better performance.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Analysis of Speech from People with Parkinson’s Disease through Nonlinear Dynamics

Orozco-Arroyave

Arias-Londoño

Vargas-Bonilla

et al. 2013

Advances in Nonlinear Speech Processing

View full text Add to dashboard Cite

show abstract

“…This result can also be noted in the receiver operating characteristic (ROC) curves that are shown in figure 3. These kind of curves are widely used in clinical applications and the area under such curves (AUC) is considered as a good statistic for representing the general performance of the system [19] …”

Section: Resultsmentioning

confidence: 99%

“…Table 2. Index allocation for features Mean 1 2 3 4 5 6 7 8 9 10 11 12 13 Std 14 15 16 17 18 19 20 21 22 23 24 25 26 Kurtosis 27 28 29 30 31 32 33 34 35 36 37 38 39 Skewness 40 41 42 43 44 45 46 47 48 49 50 The tests performed over the proposed system have been made following the strategy indicated in [19]. The 70% of the data are used for the feature selection and for training the classifier and the remaining 30% is for testing; ten different subsets for training and testing are randomly formed in ten repetitions of the randomization of the data, in order to perform a total of ten independent experiments, each one with its results, allowing the calculation of confidence intervals for the general performance analysis and robustness of the proposed system.…”

Section: Experimental Frameworkmentioning

confidence: 99%

New Cues in Low-Frequency of Speech for Automatic Detection of Parkinson’s Disease

Belalcázar-Bolaños

Orozco-Arroyave

Vargas-Bonilla

et al. 2013

Natural and Artificial Models in Computation and Biology

View full text Add to dashboard Cite

Abstract. In this paper, the analysis of low-frequency zone of the speech signals from the five Spanish vowels, by means of the Teager energy operator (TEO) and the modified group delay functions (MGDF) is proposed for the automatic detection of Parkinson's disease.According to our findings, different implementations of the TEO are suitable for tackling the problem of the automatic detection of Parkinson's disease. Additionally, the application of MGDF for improving the resolution of the speech spectrum in the low frequency zone is able for enhancing differences exhibited between the first two formants from speech of people with Parkinson's disease and healthy controls.The best results are obtained for vowel /e/, where accuracy rates of up to 92% are achieved. This is an interesting result specially if it is considered that there are muscles that are involved in the production of the vowel /e/ but not in other vowels.

show abstract

“…A subset of 173 pathological and 53 normal registers has been taken, according to those enumerated by Parsa et al [2]. The asymmetry in the amount of normal and pathological records has not been considered a problem due to the fact that pathological recordings are approximately 1 s long, whereas normal recordings last around 3 s. For a more detailed discussion of this database, see [11].…”

Section: A Databasementioning

confidence: 99%

Effects of Audio Compression in Automatic Detection of Voice Pathologies

Sáenz-Lechón

Osma-Ruiz

Godino-Llorente

et al. 2008

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

] T. Hildebrand and P. Rüegsegger, "A new method for the modelindependent assessment of thickness in three-dimensional images," J. Microsc., vol. 185, pp. 67-75, Jan. 1997 Effects of Audio Compression in Automatic Detection of Voice PathologiesNicolás Sáenz-Lechón * , Víctor Osma-Ruiz, Juan I. Godino-Llorente, Manuel Blanco-Velasco, Fernando Cruz-Roldán, and Julián D. Arias-LondoñoAbstract-This paper investigates the performance of an automatic system for voice pathology detection when the voice samples have been compressed in MP3 format and different binary rates (160, 96, 64, 48, 24, and 8 kb/s). The detectors employ cepstral and noise measurements, along with their derivatives, to characterize the voice signals. The classification is performed using Gaussian mixtures models and support vector machines. The results between the different proposed detectors are compared by means of detector error tradeoff (DET) and receiver operating characteristic (ROC) curves, concluding that there are no significant differences in the performance of the detector when the binary rates of the compressed data are above 64 kb/s. This has useful applications in telemedicine, reducing the storage space of voice recordings or transmitting them over narrow-band communications channels.Index Terms-Gaussian mixture models, MP3 compression, support vector machines, voice pathology detection.

show abstract

Methodological issues in the development of automatic systems for voice pathology detection

Cited by 140 publications

References 26 publications

Analysis of Speech from People with Parkinson’s Disease through Nonlinear Dynamics

Analysis of Speech from People with Parkinson’s Disease through Nonlinear Dynamics

New Cues in Low-Frequency of Speech for Automatic Detection of Parkinson’s Disease

Effects of Audio Compression in Automatic Detection of Voice Pathologies

Contact Info

Product

Resources

About