Reconhecimento de comandos de voz com e sem disartria usando extração de características da fala MFCC e algoritmos de aprendizagem de máquina

• Jordana Seixas UPE
• Ailton Leite UPE
• Rodrigo de Paula Unicap
• Sérgio Murilo Maciel Fernandes UPE

DOI: https://doi.org/10.5753/sbcas.2023.229708

Resumo

A fala disártrica está entre os problemas para articular e pronunciar bem as palavras devido aos danos no sistema neurológico responsável pela fala. Este estudo investiga se os classificadores de aprendizagem de máquina reconhecem quais palavras as pessoas com e sem disartria falam, aplicando uma técnica de extração de características da fala chamada MFCC (Mel Frequency Cepstral Coefficients). Os classificadores Artificial Neural Network (ANN), Support Vector Machine (SVM), Random Forest (RF) e KNearest Neighbor (KNN) foram testados. O conjunto de dados UASpeech foi usado nos modelos, contendo falantes com e sem disartria. Os resultados mostraram bom desempenho com acurácia média para KNN (98,5%), ANN (95%), RF (91,8%) e SVM (89,5%).

Referências

Boualoulou, N., Nsiri, B., Drissi, T.B. and Zayrit, S., 2022. Speech analysis for the detection of Parkinson’s disease by combined use of empirical mode decomposition, Mel frequency cepstral coefficients, and the K-nearest neighbor classifier. In ITM Web of Conferences (Vol. 43, p. 01019). EDP Sciences.

Harris-Birtill, D. and Harris-Birtill, R., 2021. Understanding computation time: a critical discussion of time as a computational performance metric. In Time in Variance (pp. 220-248). Brill.

Hashimoto, BF, Shigueoka, LS, Costa, VP e Gomi, ES, 2021, junho. Treinamento de Classificadores de Aprendizagem de Máquina para o Diagnóstico do Glaucoma com o Uso de Dados de Perimetria Automatizada Padrão (SAP). In Anais Estendidos do XXI Simpósio Brasileiro de Computação Aplicada à Saúde (pp. 151-156). SBC.

Joshy, A. A., & Rajan, R. (2022). Automated Dysarthria Severity Classification: A Study on Acoustic Features and Deep Learning Techniques. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 30, 1147-1157.

Kim, H., Hasegawa-Johnson, M., Perlman, A., Gunderson, J., Huang, T. S., Watkin, K., & Frame, S. (2008). Dysarthric speech database for universal access research. In Ninth Annual Conference of the International Speech Communication Association.

Kuresan, H., Samiappan, D., Jeevan, A. and Gupta, S., 2021. A Performance Study of ML Models and Neural Networks for Detection of Parkinson Disease using Dysarthria Symptoms. European Journal of Molecular & Clinical Medicine, 8(03), p.2021.

Mooijman, P., Catal, C., Tekinerdogan, B., Lommen, A. and Blokland, M., 2023. The effects of data balancing approaches: A case study. Applied Soft Computing, 132, p.109853.

Sahane, P., Pangaonkar, S., & Khandekar, S. (2021, September). Dysarthric Speech Recognition using Multi-Taper Mel Frequency Cepstrum Coefficients. In 2021 International Conference on Computing, Communication and Green Engineering (CCGE) (pp. 1-4). IEEE.

Sheth, V., Tripathi, U. and Sharma, A., 2022. A Comparative Analysis of Machine Learning Algorithms for Classification Purpose. Procedia Computer Science, 215, pp.422-431.

Trentin, E. and Gori, M., 2001. A survey of hybrid ANN/HMM models for automatic speech recognition. Neurocomputing, 37(1-4), pp.91-126.

Yakoub, M.S., Selouani, S.A., Zaidi, B.F. and Bouchair, A., 2020. Improving dysarthric speech recognition using empirical mode decomposition and convolutional neural network. EURASIP Journal on Audio, Speech, and Music Processing, 2020(1), pp.17.

Yılmaz, E., Mitra, V., Sivaraman, G. and Franco, H., 2019. Articulatory and bottleneck features for speaker-independent ASR of dysarthric speech. Computer Speech & Language, 58, pp.319-334.