Classification of Voice Disorders Using a One-Dimensional Convolutional Neural Network

Fujimura, Shintaro; Kojima, Tsuyoshi; Okanoue, Yusuke; Shoji, Kazuhiko; Inoue, Masato; Omori, Koichi; Hori, Ryusuke

doi:10.1016/j.jvoice.2020.02.009

Cited by 30 publications

(11 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Our iPhone application offers all the advantages of subjective evaluation without the weaknesses. Previously, we used a deep NN architecture based on TensorFlow for pathological voice evaluation using an auditory‐perceptual scale 2 . We then suggested that an end‐to‐end one‐dimensional CNN could objectively and effectively classify pathological voice severity based on the GRBAS scale.…”

Section: Discussionmentioning

confidence: 99%

“…Voice disorders are caused by vocal‐cord polyps, nodules, cysts, atrophy, paralysis, and cancer, as well as by laryngitis. Voices were recorded in a soundproof room as 16‐bit/48‐kHz WAV files, as described previously 2 . Three experts rated each sample using the GRBAS scale and median values were calculated.…”

Section: Methodsmentioning

confidence: 99%

“…The grade, roughness, breathiness, asthenia, and strain (GRBAS) scale is widely used for perceptual evaluation of voice quality 1 . We previously suggested that artificial intelligence (AI) could objectively and effectively classify pathological voice data using this scale 2 . The deep‐learning architecture for voice quality classification is based on TensorFlow, which is a software library that was developed and released by Google (Mountain View, CA, USA) for use in deep learning.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A New iPhone Application for Voice Quality Assessment Based on the GRBAS Scale

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A New iPhone Application for Voice Quality Assessment Based on the GRBAS Scale

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Assim, é neste contexto da quantificação da qualidade de vozes patológicas que as análises perceptivas-auditivas da voz permanecem como método padrão nos diagnósticos de pacientes com distúrbios vocais. (FUJIMURA et al, 2020). De modo que, a categoria dos professores apresenta alto índice de disfunção vocal, uma vez que formam um grande grupo de usuários profissionais da voz e acredita-se que correm maior risco relacionados a problemas vocais do que a população em geral (RUSSELL, OATES, GREENWOOD, 1997).…”

Section: Introductionunclassified

“…A avaliação perceptivo-auditiva da voz é uma habilidade clínica subjetiva e, portanto, podem variar entre os examinadores. Embora muitas métricas acústicas tenham sido estudadas para potencializar a avaliação objetiva de vozes patológicas, a interpretação das mesmas, em casos individuais se torna difícil, e a técnica não é amplamente utilizada pelos médicos, que acabam confiando em diretrizes e protocolos para ajudá-los a julgar a qualidade da voz de seus pacientes (FUJIMURA et al, 2020).…”

Section: Introductionunclassified

Estimação Inteligente do Grau Geral de Desvio Vocal Utilizando Transformada Wavelet Packet

Barizão

Yoshizumi

Spatti

et al. 2020

Anais Do Congresso Brasileiro De Automática 2020

View full text Add to dashboard Cite

O atual modelo de avaliação perceptivo-auditiva da voz está alicerçado em escalas analógicas visuais e buscam melhorias contínuas sob inúmeros aspectos, com destaque para a prescrição de uma boa avaliação referente a qualidade vocal, bem como o seu diagnóstico e monitoramento de tratamentos. Refletindo-se sobre as eventualidades que degradam os aspectos, constata-se a não padronização dos protocolos avaliativos comumente utilizados. Assim sendo, os objetivos principais deste artigo consistiram em desenvolver ferramentas computacionais dedicadas à estimação do parâmetro subjetivo, grau geral do desvio vocal. Mais especificamente, desenvolveu-se a técnica de fragmentação dos arquivos de áudios, a decomposição downsample desses dados para o nível seis dos coeficientes de energia da transformada wavelet packet referentes a seis filtros wavelets distintos, sendo eles: daubechies, symlets, coiflets, fejer-korovkin, biorthogonal e reverse biorthogonal, propondo uma metodologia automatizada para implementação das redes neurais perceptron multicamadas com a finalidade de integrar as variáveis advindas de tal decomposição. Resultados com amostras vocais reais são apresentados para demonstrar a eficiência das técnicas desenvolvidas, expondo taxas de acertos maiores que 98.7% para o método elaborado.

show abstract

The Effect of Noise on Deep Learning for Classification of Pathological Voice

Hasebe,

Fujimura,

Kojima

et al. 2024

The Laryngoscope

Self Cite

View full text Add to dashboard Cite

ObjectiveThis study aimed to evaluate the significance of background noise in machine learning models assessing the GRBAS scale for voice disorders.MethodsA dataset of 1406 voice samples was collected from retrospective data, and a 5‐layer 1D convolutional neural network (CNN) model was constructed using TensorFlow. The dataset was divided into training, validation, and test data. Gaussian noise was added to test samples at various intensities to assess the model's noise resilience. The model's performance was evaluated using accuracy, F1 score, and quadratic weighted Cohen's kappa score.ResultsThe model's performance on the GRBAS scale generally declined with increasing noise intensities. For the G scale, accuracy dropped from 70.9% (original) to 8.5% (at the highest noise), F1 score from 69.2% to 1.3%, and Cohen's kappa from 0.679 to 0.0. Similar declines were observed for the remaining RBAS components.ConclusionThe model's performance was affected by background noise, with substantial decreases in evaluation metrics as noise levels intensified. Future research should explore noise‐tolerant techniques, such as data augmentation, to improve the model's noise resilience in real‐world settings.Level of EvidenceThis study evaluates a machine learning model using a single dataset without comparative controls. Given its non‐comparative design and specific focus, it aligns with Level 4 evidence (Case‐series) under the 2011 OCEBM guidelines Laryngoscope, 2024

show abstract

Classification of Voice Disorders Using a One-Dimensional Convolutional Neural Network

Cited by 30 publications

References 10 publications

A New iPhone Application for Voice Quality Assessment Based on the GRBAS Scale

A New iPhone Application for Voice Quality Assessment Based on the GRBAS Scale

Estimação Inteligente do Grau Geral de Desvio Vocal Utilizando Transformada Wavelet Packet

The Effect of Noise on Deep Learning for Classification of Pathological Voice

Contact Info

Product

Resources

About