Automatic dysarthria evaluation system and method based on speech recognition

Abstract

The invention provides an automatic dysarthria evaluation system and method based on speech recognition. The system comprises a first feature extraction unit, a second feature extraction unit, a feature splicing unit, a multi-layer perceptron and an evaluation unit. The feature splicing unit is in communication connection with the first feature extraction unit, the second feature extraction unit and the multi-layer perceptron, and the evaluation unit is in communication connection with the multi-layer perceptron. The first feature extraction unit is used for extracting traditional sentence-level acoustic features. The second feature extraction unit extracts frame-level audio annotations and the frame phoneme-probability relationship. The feature splicing unit is used for splicing the features extracted by the first feature extraction unit and the features extracted by the second feature extraction unit. The multi-layer perceptron outputs the individual sentence obstacle degree and the corresponding prediction probability based on the spliced features. The evaluation unit obtains an overall evaluation result by using the prediction probability information of the individual sentences. According to the invention, the accuracy and stability of dysarthria estimation can be improved.

Description

technical field [0001] The invention relates to the technical field of dysarthria assessment, in particular to an automatic assessment system and method for dysarthria based on speech recognition. Background technique [0002] Dysarthria manifests as slurred speech, fluent speech, inaccurate pronunciation, and abnormal volume and rhythm. Doctors usually check whether there is dysarthria and the degree of pathology through examination of vocal organs and speech assessment. For preschool children, the above performance can be improved and cured through language training. Physician resources, time constraints, and the widespread use of the Internet and mobile devices have prompted language training for dysarthria to be delivered on mobile device applications (apps). The evaluation results of language training effects on the mobile terminal can provide users with timely feedback, and at the same time provide training program designers with important information that is conduci...

AI Technical Summary

Problems solved by technology

For consonants, this method only focuses on the six consonants b, p, d, t, g, and k. At the same time, it is difficult to achieve accurate automatic extraction of features such as voice onset time

In addition, these features are not enough to fully reflect the problems of dysarthria, especially the inaccurate pronunciation caused by the substitution of consonants.

[0005] To sum up, the existing technology still lacks an effective automatic assessment method for dysarthria, and the main problems are: the subjective assessment method of auditory perception lacks objectivity, accuracy and stability; the automatic assessment of dysarthria has not been realized; the existing assessment The input used by the method is limited to limited, isolated letter sounds, and no continuous speech information is used

Images