Automatic Detection of Pathological Voices Using Complexity Measures, Noise Parameters, and Mel-Cepstral Coefficients

Arias-Londoño, Julián D.; Godino-Llorente, Juan Ignacio; Sáenz-Lechón, Nicolás; Osma-Ruiz, Víctor; Castellanos-Domínguez, Germán

doi:10.1109/tbme.2010.2089052

Cited by 131 publications

(83 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The recordings are preprocessed by means of its division into frames with 55ms of length with an overlap of 50%, according to [9]. After, NLD features are calculated for each frame and four statistics are calculated for each feature.…”

Section: Methodsmentioning

confidence: 99%

“…In this paper, different state of the art NLD features are implemented and their discrimination capability is objectively evaluated on the automatic classification of speech signals from PPD and CS. The set of features considered in this study includes a total of 10 measures which have been used for the automatic detection of different speech disorders such as hypernasality [8] and dysphonia [5], [9]. The features are: correlation dimension, largest Lyapunov exponent, Lempel-Ziv complexity, Hurst exponent, RPDE, DFA, approximate entropy, approximate entropy with Gaussian kernel, sample entropy, sample entropy with Gaussian kernel.…”

Section: Introductionmentioning

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: Introductionmentioning

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

“…Changes in spectral envelope shape and slope are widely reported as a consequence of voice pathology [13] or Parkinson's [14,15]. The use of mel-frequency cepstral coefficients (MFCC) to describe the short-time speech spectral envelope is very common.…”

Section: Features For Detecting Health Changesmentioning

confidence: 99%

It Sounds Like You Have a Cold! Testing Voice Features for the Interspeech 2017 Computational Paralinguistics Cold Challenge

Huckvale

Beke

2017

Interspeech 2017

View full text Add to dashboard Cite

This paper describes an evaluation of four different voice feature sets for detecting symptoms of the common cold in speech as part of the Interspeech 2017 Computational Paralinguistics Challenge. The challenge corpus consists of 630 speakers in three partitions, of which approximately one third had a "severe" cold at the time of recording. Success on the task is measured in terms of unweighted average recall of cold/not-cold classification from short extracts of the recordings. In this paper we review previous voice features used for studying changes in health and devise four basic types of features for evaluation: voice quality features, vowel spectra features, modulation spectra features, and spectral distribution features. The evaluation shows that each feature set provides some useful information to the task, with features from the modulation spectrogram being most effective. Feature-level fusion of the feature sets shows small performance improvements on the development test set. We discuss the results in terms of the most suitable features for detecting symptoms of cold and address issues arising from the design of the challenge.

show abstract

“…From past decade mel frequency cepstral coefficients (MFCCs) have been widely used in pathological voice detection systems with good results. [8][9][10] The main advantage of MFCCs is that, its calculation does not require any previous pitch information, which is a difficult task in the presence of pathology. The basic time domain features and MFCCs can be used in developing classifiers models for detection of pathological voices.…”

Section: Introductionmentioning

confidence: 99%

A hybrid model for neurological disordered voice classification using time and frequency domain features

K¹,

Holi

2015

AIR

View full text Add to dashboard Cite

Different neurological disorders may lead to speech related problems, due to paralysis in vocal fold or weakness of the related muscles. This may modify the acoustic characteristics of the subject's voice which may provide important information for detecting certain neurological diseases. The vowel phonation which is acoustically informative and uttered by the patient with not much difficulty is collected and various acoustic features are extracted by time domain and frequency domain techniques. The use of all these features for classification may lead to a large feature space, which may lead to complexity. Hence to avoid this, in the present work experimentation is done by fusing different classifiers which are fed with features extracted from different domains. The time domain and frequency domain features are given to Support Vector Machine (SVM) and Gaussian Mixture Model (GMM) respectively, and the intermediate decision of these classifiers is given to another SVM to identify the voice signal as normal or diseased. It is observed that this hybrid classifier model has shown some improvement with a classification accuracy of 91.43% compared to single GMM classifier with an accuracy of classification of 90% with frequency domain features as input.

show abstract

Automatic Detection of Pathological Voices Using Complexity Measures, Noise Parameters, and Mel-Cepstral Coefficients

Cited by 131 publications

References 43 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

It Sounds Like You Have a Cold! Testing Voice Features for the Interspeech 2017 Computational Paralinguistics Cold Challenge

A hybrid model for neurological disordered voice classification using time and frequency domain features

Contact Info

Product

Resources

About