Small-size spectral features for machine learning in voice signal analysis and classification tasks

Objectives. The problem of developing a method for calculating small-sized spectral features that increases the efficiency of existing machine learning systems for analyzing and classifying voice signals is being solved.

Methods. Spectral features are extracted using a generative approach, which involves calculating a discrete Fourier spectrum for a sequence of samples generated using an autoregressive model of input voice signal. The generated sequence processed by the discrete Fourier transform considers the periodicity of the transform and thereby increase the accuracy of spectral estimation of analyzed signal.

Results. A generative method for calculating spectral features intended for use in machine learning systems for the analysis and classification of voice signals is proposed and described. An experimental analysis of the  accuracy and stability of the spectrum representation of a test signal with a known spectral composition has been carried out using the envelopes. The envelopes were calculated using  proposed generative method and using discrete Fourier transform with different analysis windows (rectangular window and Hanna window).  The analysis showed that spectral envelopes obtained using the proposed method more accurately represent the spectrum of test signal according to the criterion of minimum square error. A comparison of the effectiveness of voice signal classification with proposed features and the features based on the mel-frequency kepstral  coefficients is carried out. A diagnostic system for amyotrophic lateral sclerosis was used as a basic test system to evaluate the effectiveness of proposed approach in practice. 

Conclusion. The obtained experimental results showed a significant increase of classification accuracy when using proposed approach for calculating features compared with the features based on the mel-frequency kepstral coefficients.

1. Harar P., Galaz Z., Alonso-Hernandez J. B., Mekyska J., Burget R., Smekal Z. Towards robust voice pathology detection. Neural Computing and Applications, 2020, vol. 32, no. 20, pp. 15747–15757.

2. Benba A., Jilbab A., Hammouch A. Discriminating between patients with Parkinson’s and neurological diseases using cepstral analysis. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2016, vol. 24, no. 10, pp. 1100–1108.

3. Vashkevich M., Rushkevich Y. Classification of ALS patients based on acoustic analysis of sustained vowel phonations. Biomedical Signal Processing and Control, 2021, vol. 65, pp. 1–14.

4. Rabiner L. R., Juang B. H. Fundamentals of Speech Recognition. Pearson Education, 1993, 570 p.

5. Harris F. J. On the use of windows for harmonic analysis with the discrete Fourier transform. Proceedings of the IEEE, January 1978, vol. 66, no. 1, pp. 51–83. https://doi.org/10.1109/PROC.1978.10837

6. Vashkevich M. I., Likhachov D. S., Azarov E. S. Voice analysis and classification system based on perturbation parameters and cepstral presentation in psychoacoustic scales. Doklady Belorusskogo gosudarstvennogo universiteta informatiki i radioèlektroniki [Reports of the Belarusian State University of Informatics and Radioelectronics], 2022, vol. 20, no. 1, pp. 73–82 (In Russ.). https://doi.org/10.35596/17297648-2022-20-1-73-82

7. Markel J. D., Gray A. H. Linear Prediction of Speech. Berlin, New York, Springer-Verlag, 1976, 290 p.

8. Flach P. Machine Learning: The Art and Science of Algorithms That Make Sense of Data. Cambridge University Press, 2012, 416 p.

9. James G., Witten D., Hastie T., Tibshirani R. An Introduction to Statistical Learning with Applications in R. Springer, 2013, 440 p.

10. Vashkevich, M., Petrovsky A., Rushkevich Y. Bulbar ALS detection based on analysis of voice perturbation and vibrato. IEEE International Conference on Signal Processing: Algorithms, Architectures, Arrangements, and Applications, Poznan, Poland, 18–20 September 2019. Poznan, 2019, pp. 267–272.

11. Kunjan S., Grummett T. S., Pope K. J., Powers D. M. W., Fitzgibbon S. P., …, Lewis T. W. The necessity of leave one subject out (LOSO) cross validation for EEG disease diagnosis. Brain Informatics, Springer, 2021, pp. 558–567. https://doi.org/10.1007/978-3-030-86993-9_50