Rapid dysarthria detection method based on modal decomposition

Abstract

The invention relates to a rapid dysarthria detection method based on modal decomposition. The method comprises the following steps: collecting an original voice signal; preprocessing the original voice signal; performing acoustic feature extraction on the framed and windowed signal S based on modal decomposition to obtain statistical features; the statistical features are input into a machine learning classifier, dysarthria detection is achieved, and the machine learning classifier is a support vector machine (SVM) model. According to the method, the limitation of traditional acoustic features in a nonlinear time-varying system is overcome, the IMF obtained through decomposition contains time-frequency information of original audio signals on different levels, voice physiological information of dysarthria patients can be well captured, pathological changes of vocal organs are reflected, and the accuracy of the voice physiological information is improved. The accuracy and robustness of dysarthria detection are improved; the method can adapt to non-linear and non-stable voice signals, and further improves the detection effect of dysarthria.

Description

technical field [0001] The invention relates to the technical field of dysarthria detection, in particular to a method for rapid detection of dysarthria based on modal decomposition. Background technique [0002] Dysarthria is abnormal breathing, vocalization, pronunciation, resonance and rhythm caused by pathological changes of speech organs or nervous system, abnormal morphology, manifested as dysphonia, inaccurate pronunciation, slurred articulation, sound, pitch and rate, rhythm and other abnormalities , as well as changes in speech and auditory characteristics such as nasal hyperintensity. It is clinically common in neurological diseases such as Parkinson's disease, which seriously affects the quality of life and social life of patients. [0003] At present, there is no specific evaluation standard for dysarthria, and most of the clinical subjective methods of auditory perception are used, such as Frenchay evaluation method. This needs to be examined, recorded and sco...

AI Technical Summary

Problems solved by technology

However, these features do not take into account the time-frequency characteristics of sound

In this regard, some other studies are based on spectral and cepstrum features. However, these features can only reflect the static characteristics of the speech signal, and there are limitations in directly processing nonlinear and non-stationary speech signals, ignoring the implicit information of the sound, resulting in detection Not accurate enough, less robust

Images