Vinicius Jefferson Dias Vieira obtained his Ph.D. degree at the Electrical Engineering Department from the Federal University of Campina Grande (UFCG) in 2018. He obtained the M.Sc. degree from the Federal Institute of Education, Science and Technology of Paraiba (IFPB) in 2014. From the IFPB, Vinicius is also graduated at the course of Technology in Telecommunications Systems (2012). Member of the Laboratory of Acoustic Signal Processing (LASP), at Military Institute of Engineering (IME). Member of the Brazilian Telecommunications Society (SBrT). Currently, his interest fields are: digital speech signal processing, time-frequency acoustic analysis, nonlinear dynamic systems and pattern recognition.
Tecnicas de processamento digital de sinais tem sido fortemente utilizadas atraves da analise acu... more Tecnicas de processamento digital de sinais tem sido fortemente utilizadas atraves da analise acustica de desordens provocadas por patologias laringeas, devido a sua simplicidade e natureza nao invasiva. No reconhecimento de padroes, uma tecnica para classificacao de sinais, que vem sendo utilizada recentemente, devido a sua robustez diante de dados com grande dimensao e boa capacidade de generalizacao, sao as maquinas de vetores de suporte. Neste artigo e apresentada uma aplicacao desta tecnica para a classificacao de sinais de vozes saudaveis e vozes afetadas por patologias na laringe, especificamente, edema de Reinke, paralisia nas pregas vocais e nodulos. Foram utilizadas quatro maquinas de vetor de suporte, uma para cada classe de sinais. A caracteristica fornecida ao classificador, no processo de treinamento, e o parâmetro de Hurst obtido pelo metodo da variância no tempo. Os resultados apresentaram taxa de correta classificacao superior a 99% na discriminacao entre os sinais ...
Este trabalho trata da analise acustica de sinais de vozes saudaveis e de vozes afetadas por para... more Este trabalho trata da analise acustica de sinais de vozes saudaveis e de vozes afetadas por paralisia nas pregas vocais, utilizando a tecnica de quantificacao de recorrencia. E investigado o melhor valor para o parâmetro que define os pontos recorrentes dentro do grafico de recorrencia, denominado raio de vizinhanca, no qual se obtenha as melhores taxas de classificacao. As medidas de quantificacao de recorrencia empregadas sao: Taxa de Recorrencia, Determinismo, Comprimento maximo das linhas diagonais, Entropia de Shannon, Tendencia, Laminaridade, Tempo de permanencia em um estado e Comprimento maximo das linhas verticais. E avaliado o desempenho das medidas de forma individual e combinada. As combinacoes sao realizadas, considerando o valor de raio de vizinhanca que obtem melhor acuracia individual para cada medida. A classificacao e realizada por meio da analise discriminante, com as funcoes linear e quadratica. O melhor resultado obtido por meio de validacao cruzada indica uma...
Studies in health technology and informatics, 2015
This paper deals with the discrimination between healthy and pathological speech signals using re... more This paper deals with the discrimination between healthy and pathological speech signals using recurrence plots and wavelet transform with texture features. Approximation and detail coefficients are obtained from the recurrence plots using Haar wavelet transform, considering one decomposition level. The considered laryngeal pathologies are: paralysis, Reinke's edema and nodules. Accuracy rates above 86% were obtained by means of the employed method.
This work summarizes the research related to digital speech signal processing with recurrence qua... more This work summarizes the research related to digital speech signal processing with recurrence quantification analysis (RQA) applied to voice disorder assessment. The main motivation for these studies is the fact that RQA is able to exploit the nonlinear dynamical nature of the speech production system. Due to the use of recurrence quantification measures to represent the behavior of speech signals, promising results were obtained in the characterization and classification of laryngeal pathologies and voice disorders. These contributions may help one to evaluate the usability and efficiency of RQA in vocal disorder assessment.
In this paper, the performance of quantification measures of recurrence plots is evaluated in the... more In this paper, the performance of quantification measures of recurrence plots is evaluated in the task of discriminating pathological voices from the healthy ones. In order to classify these signals as healthy, edema or paralysis, seven recurrence quantification measures are used: Determinism (DET), Maximum length of the diagonal structures (Lmax), Entropy (ENTR), Slope of line-of-best-fit (TREND), Laminarity (LAM), Length of
Este artigo trata da aplicação da análise de
quantificação de recorrência a curto e a longo inter... more Este artigo trata da aplicação da análise de quantificação de recorrência a curto e a longo intervalo de tempo em sinais de voz de laringes saudáveis e sinais de voz de laringes patológicas (paralisia, edema e nódulos). O objetivo é identificar qual medida (ou conjunto de medidas) e qual das formas de análise apresenta melhor desempenho. São extraídas, a partir dos gráficos de recorrência, seis medidas de quantificação: determinismo, comprimento máximo das linhas diagonais, entropia de Shannon, laminaridade, tempo de permanência e comprimento máximo das linhas verticais. Os resultados obtidos, empregando um classificador baseado em análise discriminante quadrática, indicam que a análise a longo intervalo de tempo é mais promissora, principalmente com a medida do comprimento máximo das linhas diagonais, com a qual se atinge uma acurácia de até 96,27%±1,53%.
Palavras-chave: Análise de quantificação de recorrência, processamento digital de sinais de voz, patologias laríngeas.
Patologias na laringe causam distúrbios na voz que podem ser detectados por meio de técnicas de p... more Patologias na laringe causam distúrbios na voz que podem ser detectados por meio de técnicas de processamento digital de sinais. A análise acústica desses sinais, comparativamente a sinais de vozes produzidos por locutores com laringes saudáveis, pode ser empregada como uma ferramenta de apoio ao diagnóstico de patologias laríngeas, bem como ao tratamento terapêutico de disfonias e acompanhamento pré e pós-cirúrgicos. A eficiência do método depende de fatores tais como a escolha das características ou parâmetros que melhor representem a patologia ou o distúrbio vocal, bem como do método de classificação empregado. Este artigo apresenta alguns métodos baseados no modelo linear de produção da fala, como também na análise dinâmica não linear para a classificação de patologias na laringe.
Palavras-chave: patologias na laringe, análise acústica, análise linear, análise não-linear.
Abstract: Objective: To analyze the accuracy of recurrence measurements, both isolated and combin... more Abstract: Objective: To analyze the accuracy of recurrence measurements, both isolated and combined, to assess the intensity of vocal disorders in children. Method: A total of 93 children of both sexes (48 girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the visual-analog scale. In the acoustic analysis, eight recurrence-plot characteristics were evaluated, extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%, and 5%. The classification was performed using quadratic discriminant analysis applied for individual and combined measurements. The performance was evaluated by measuring the accuracy, which related the cases correctly classified to all the analyzed cases. Results: In the classification cases concerning individual measure performance, the trapping time and maximum length of the diagonal lines showed the best classification potential to discriminate between healthy and disturbed voices, with accuracy rates above 80%. In the healthy and mild-deviation cases, the TREND measure was also relevant. For the mild- versus moderate-deviation classification, the best performance was obtained by the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating between healthy voices and those with mild deviation. Conclusions: The measures of recurrence, either alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the intensity of vocal disorders in children.
Objective
To analyze the accuracy of recurrence measurements, both isolated and combined, to asse... more Objective To analyze the accuracy of recurrence measurements, both isolated and combined, to assess the intensity of vocal disorders in children.
Method A total of 93 children of both sexes (48 girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the visual analog scale. In the acoustic analysis, eight recurrence plot characteristics were evaluated and extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%, and 5%. The classification was performed using quadratic discriminant analysis applied for individual and combined measurements. The performance was evaluated by measuring the accuracy, which related the cases correctly classified to all the analyzed cases.
Results In the classification cases concerning individual measure performance, the trapping time and maximum length of the diagonal lines showed the best classification potential to discriminate between healthy and disturbed voices, with accuracy rates above 80%. In the healthy and mild deviation cases, the trend (TREND) measure was also relevant. For the mild versus moderate deviation classification, the best performance was obtained by the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating between healthy voices and those with mild deviation.
Conclusions The measures of recurrence, either alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the intensity of vocal disorders in children.
In this paper, the performance of quantification measures of recurrence plots is evaluated in the... more In this paper, the performance of quantification measures of recurrence plots is evaluated in the task of discriminating pathological voices from the healthy ones. In order to classify these signals as healthy, edema or paralysis, seven recurrence quantification measures are used: Determinism (DET), Maximum length of the diagonal structures (Lmax), Entropy (ENTR), Slope of line-of-best-fit (TREND), Laminarity (LAM), Length of
Tecnicas de processamento digital de sinais tem sido fortemente utilizadas atraves da analise acu... more Tecnicas de processamento digital de sinais tem sido fortemente utilizadas atraves da analise acustica de desordens provocadas por patologias laringeas, devido a sua simplicidade e natureza nao invasiva. No reconhecimento de padroes, uma tecnica para classificacao de sinais, que vem sendo utilizada recentemente, devido a sua robustez diante de dados com grande dimensao e boa capacidade de generalizacao, sao as maquinas de vetores de suporte. Neste artigo e apresentada uma aplicacao desta tecnica para a classificacao de sinais de vozes saudaveis e vozes afetadas por patologias na laringe, especificamente, edema de Reinke, paralisia nas pregas vocais e nodulos. Foram utilizadas quatro maquinas de vetor de suporte, uma para cada classe de sinais. A caracteristica fornecida ao classificador, no processo de treinamento, e o parâmetro de Hurst obtido pelo metodo da variância no tempo. Os resultados apresentaram taxa de correta classificacao superior a 99% na discriminacao entre os sinais ...
Este trabalho trata da analise acustica de sinais de vozes saudaveis e de vozes afetadas por para... more Este trabalho trata da analise acustica de sinais de vozes saudaveis e de vozes afetadas por paralisia nas pregas vocais, utilizando a tecnica de quantificacao de recorrencia. E investigado o melhor valor para o parâmetro que define os pontos recorrentes dentro do grafico de recorrencia, denominado raio de vizinhanca, no qual se obtenha as melhores taxas de classificacao. As medidas de quantificacao de recorrencia empregadas sao: Taxa de Recorrencia, Determinismo, Comprimento maximo das linhas diagonais, Entropia de Shannon, Tendencia, Laminaridade, Tempo de permanencia em um estado e Comprimento maximo das linhas verticais. E avaliado o desempenho das medidas de forma individual e combinada. As combinacoes sao realizadas, considerando o valor de raio de vizinhanca que obtem melhor acuracia individual para cada medida. A classificacao e realizada por meio da analise discriminante, com as funcoes linear e quadratica. O melhor resultado obtido por meio de validacao cruzada indica uma...
Studies in health technology and informatics, 2015
This paper deals with the discrimination between healthy and pathological speech signals using re... more This paper deals with the discrimination between healthy and pathological speech signals using recurrence plots and wavelet transform with texture features. Approximation and detail coefficients are obtained from the recurrence plots using Haar wavelet transform, considering one decomposition level. The considered laryngeal pathologies are: paralysis, Reinke's edema and nodules. Accuracy rates above 86% were obtained by means of the employed method.
This work summarizes the research related to digital speech signal processing with recurrence qua... more This work summarizes the research related to digital speech signal processing with recurrence quantification analysis (RQA) applied to voice disorder assessment. The main motivation for these studies is the fact that RQA is able to exploit the nonlinear dynamical nature of the speech production system. Due to the use of recurrence quantification measures to represent the behavior of speech signals, promising results were obtained in the characterization and classification of laryngeal pathologies and voice disorders. These contributions may help one to evaluate the usability and efficiency of RQA in vocal disorder assessment.
In this paper, the performance of quantification measures of recurrence plots is evaluated in the... more In this paper, the performance of quantification measures of recurrence plots is evaluated in the task of discriminating pathological voices from the healthy ones. In order to classify these signals as healthy, edema or paralysis, seven recurrence quantification measures are used: Determinism (DET), Maximum length of the diagonal structures (Lmax), Entropy (ENTR), Slope of line-of-best-fit (TREND), Laminarity (LAM), Length of
Este artigo trata da aplicação da análise de
quantificação de recorrência a curto e a longo inter... more Este artigo trata da aplicação da análise de quantificação de recorrência a curto e a longo intervalo de tempo em sinais de voz de laringes saudáveis e sinais de voz de laringes patológicas (paralisia, edema e nódulos). O objetivo é identificar qual medida (ou conjunto de medidas) e qual das formas de análise apresenta melhor desempenho. São extraídas, a partir dos gráficos de recorrência, seis medidas de quantificação: determinismo, comprimento máximo das linhas diagonais, entropia de Shannon, laminaridade, tempo de permanência e comprimento máximo das linhas verticais. Os resultados obtidos, empregando um classificador baseado em análise discriminante quadrática, indicam que a análise a longo intervalo de tempo é mais promissora, principalmente com a medida do comprimento máximo das linhas diagonais, com a qual se atinge uma acurácia de até 96,27%±1,53%.
Palavras-chave: Análise de quantificação de recorrência, processamento digital de sinais de voz, patologias laríngeas.
Patologias na laringe causam distúrbios na voz que podem ser detectados por meio de técnicas de p... more Patologias na laringe causam distúrbios na voz que podem ser detectados por meio de técnicas de processamento digital de sinais. A análise acústica desses sinais, comparativamente a sinais de vozes produzidos por locutores com laringes saudáveis, pode ser empregada como uma ferramenta de apoio ao diagnóstico de patologias laríngeas, bem como ao tratamento terapêutico de disfonias e acompanhamento pré e pós-cirúrgicos. A eficiência do método depende de fatores tais como a escolha das características ou parâmetros que melhor representem a patologia ou o distúrbio vocal, bem como do método de classificação empregado. Este artigo apresenta alguns métodos baseados no modelo linear de produção da fala, como também na análise dinâmica não linear para a classificação de patologias na laringe.
Palavras-chave: patologias na laringe, análise acústica, análise linear, análise não-linear.
Abstract: Objective: To analyze the accuracy of recurrence measurements, both isolated and combin... more Abstract: Objective: To analyze the accuracy of recurrence measurements, both isolated and combined, to assess the intensity of vocal disorders in children. Method: A total of 93 children of both sexes (48 girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the visual-analog scale. In the acoustic analysis, eight recurrence-plot characteristics were evaluated, extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%, and 5%. The classification was performed using quadratic discriminant analysis applied for individual and combined measurements. The performance was evaluated by measuring the accuracy, which related the cases correctly classified to all the analyzed cases. Results: In the classification cases concerning individual measure performance, the trapping time and maximum length of the diagonal lines showed the best classification potential to discriminate between healthy and disturbed voices, with accuracy rates above 80%. In the healthy and mild-deviation cases, the TREND measure was also relevant. For the mild- versus moderate-deviation classification, the best performance was obtained by the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating between healthy voices and those with mild deviation. Conclusions: The measures of recurrence, either alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the intensity of vocal disorders in children.
Objective
To analyze the accuracy of recurrence measurements, both isolated and combined, to asse... more Objective To analyze the accuracy of recurrence measurements, both isolated and combined, to assess the intensity of vocal disorders in children.
Method A total of 93 children of both sexes (48 girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the visual analog scale. In the acoustic analysis, eight recurrence plot characteristics were evaluated and extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%, and 5%. The classification was performed using quadratic discriminant analysis applied for individual and combined measurements. The performance was evaluated by measuring the accuracy, which related the cases correctly classified to all the analyzed cases.
Results In the classification cases concerning individual measure performance, the trapping time and maximum length of the diagonal lines showed the best classification potential to discriminate between healthy and disturbed voices, with accuracy rates above 80%. In the healthy and mild deviation cases, the trend (TREND) measure was also relevant. For the mild versus moderate deviation classification, the best performance was obtained by the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating between healthy voices and those with mild deviation.
Conclusions The measures of recurrence, either alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the intensity of vocal disorders in children.
In this paper, the performance of quantification measures of recurrence plots is evaluated in the... more In this paper, the performance of quantification measures of recurrence plots is evaluated in the task of discriminating pathological voices from the healthy ones. In order to classify these signals as healthy, edema or paralysis, seven recurrence quantification measures are used: Determinism (DET), Maximum length of the diagonal structures (Lmax), Entropy (ENTR), Slope of line-of-best-fit (TREND), Laminarity (LAM), Length of
Uploads
Papers by Vinícius J. D. Vieira
quantificação de recorrência a curto e a longo intervalo
de tempo em sinais de voz de laringes saudáveis e sinais
de voz de laringes patológicas (paralisia, edema e
nódulos). O objetivo é identificar qual medida (ou
conjunto de medidas) e qual das formas de análise
apresenta melhor desempenho. São extraídas, a partir dos
gráficos de recorrência, seis medidas de quantificação:
determinismo, comprimento máximo das linhas
diagonais, entropia de Shannon, laminaridade, tempo de
permanência e comprimento máximo das linhas verticais.
Os resultados obtidos, empregando um classificador
baseado em análise discriminante quadrática, indicam
que a análise a longo intervalo de tempo é mais
promissora, principalmente com a medida do
comprimento máximo das linhas diagonais, com a qual
se atinge uma acurácia de até 96,27%±1,53%.
Palavras-chave: Análise de quantificação de recorrência,
processamento digital de sinais de voz, patologias
laríngeas.
Palavras-chave: patologias na laringe, análise acústica, análise linear, análise não-linear.
to assess the intensity of vocal disorders in children. Method: A total of 93 children of both sexes (48
girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was
evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the
visual-analog scale. In the acoustic analysis, eight recurrence-plot characteristics were evaluated,
extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%,
and 5%. The classification was performed using quadratic discriminant analysis applied for individual
and combined measurements. The performance was evaluated by measuring the accuracy, which
related the cases correctly classified to all the analyzed cases. Results: In the classification cases
concerning individual measure performance, the trapping time and maximum length of the diagonal
lines showed the best classification potential to discriminate between healthy and disturbed voices,
with accuracy rates above 80%. In the healthy and mild-deviation cases, the TREND measure was also
relevant. For the mild- versus moderate-deviation classification, the best performance was obtained by
the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the
measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating
between healthy voices and those with mild deviation. Conclusions: The measures of recurrence, either
alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the
intensity of vocal disorders in children.
To analyze the accuracy of recurrence measurements, both isolated and combined, to assess the intensity of vocal disorders in children.
Method
A total of 93 children of both sexes (48 girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the visual analog scale. In the acoustic analysis, eight recurrence plot characteristics were evaluated and extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%, and 5%. The classification was performed using quadratic discriminant analysis applied for individual and combined measurements. The performance was evaluated by measuring the accuracy, which related the cases correctly classified to all the analyzed cases.
Results
In the classification cases concerning individual measure performance, the trapping time and maximum length of the diagonal lines showed the best classification potential to discriminate between healthy and disturbed voices, with accuracy rates above 80%. In the healthy and mild deviation cases, the trend (TREND) measure was also relevant. For the mild versus moderate deviation classification, the best performance was obtained by the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating between healthy voices and those with mild deviation.
Conclusions
The measures of recurrence, either alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the intensity of vocal disorders in children.
Key Words:
Voice analysis, Acoustic, Children, Nonlinear analysis
quantificação de recorrência a curto e a longo intervalo
de tempo em sinais de voz de laringes saudáveis e sinais
de voz de laringes patológicas (paralisia, edema e
nódulos). O objetivo é identificar qual medida (ou
conjunto de medidas) e qual das formas de análise
apresenta melhor desempenho. São extraídas, a partir dos
gráficos de recorrência, seis medidas de quantificação:
determinismo, comprimento máximo das linhas
diagonais, entropia de Shannon, laminaridade, tempo de
permanência e comprimento máximo das linhas verticais.
Os resultados obtidos, empregando um classificador
baseado em análise discriminante quadrática, indicam
que a análise a longo intervalo de tempo é mais
promissora, principalmente com a medida do
comprimento máximo das linhas diagonais, com a qual
se atinge uma acurácia de até 96,27%±1,53%.
Palavras-chave: Análise de quantificação de recorrência,
processamento digital de sinais de voz, patologias
laríngeas.
Palavras-chave: patologias na laringe, análise acústica, análise linear, análise não-linear.
to assess the intensity of vocal disorders in children. Method: A total of 93 children of both sexes (48
girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was
evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the
visual-analog scale. In the acoustic analysis, eight recurrence-plot characteristics were evaluated,
extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%,
and 5%. The classification was performed using quadratic discriminant analysis applied for individual
and combined measurements. The performance was evaluated by measuring the accuracy, which
related the cases correctly classified to all the analyzed cases. Results: In the classification cases
concerning individual measure performance, the trapping time and maximum length of the diagonal
lines showed the best classification potential to discriminate between healthy and disturbed voices,
with accuracy rates above 80%. In the healthy and mild-deviation cases, the TREND measure was also
relevant. For the mild- versus moderate-deviation classification, the best performance was obtained by
the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the
measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating
between healthy voices and those with mild deviation. Conclusions: The measures of recurrence, either
alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the
intensity of vocal disorders in children.
To analyze the accuracy of recurrence measurements, both isolated and combined, to assess the intensity of vocal disorders in children.
Method
A total of 93 children of both sexes (48 girls and 45 boys), aged between 3 and 10 years, participated. The vocal-deviation intensity was evaluated by the consensus of three speech therapists from the pronunciation of vowel /ε/ using the visual analog scale. In the acoustic analysis, eight recurrence plot characteristics were evaluated and extracted with neighborhood radius values that maintained the recurrence rate at 1%, 2%, 3%, 4%, and 5%. The classification was performed using quadratic discriminant analysis applied for individual and combined measurements. The performance was evaluated by measuring the accuracy, which related the cases correctly classified to all the analyzed cases.
Results
In the classification cases concerning individual measure performance, the trapping time and maximum length of the diagonal lines showed the best classification potential to discriminate between healthy and disturbed voices, with accuracy rates above 80%. In the healthy and mild deviation cases, the trend (TREND) measure was also relevant. For the mild versus moderate deviation classification, the best performance was obtained by the TREND measure (85.00% ± 7.64%). A gain was obtained in the classification rate when the measures of recurrence were combined, reaching an accuracy of 95.00% ± 5.00%, for discriminating between healthy voices and those with mild deviation.
Conclusions
The measures of recurrence, either alone or combined, may be useful in detecting healthy and disturbed voices and in differentiating the intensity of vocal disorders in children.
Key Words:
Voice analysis, Acoustic, Children, Nonlinear analysis