Classification of vocal aging using parameters extracted from the glottal signal

Area of Science:

• Speech processing
• Biomedical engineering
• Machine learning

Background:

• Vocal aging classification is crucial for understanding physiological changes.
• Existing methods often rely on standard speech signal parameters like MFCC.
• A need exists for more accurate and robust vocal aging classification techniques.

Purpose of the Study:

• To propose and evaluate a novel method for vocal aging classification.
• To compare the effectiveness of glottal signal features versus MFCC for age-related voice analysis.
• To determine the optimal feature set for classifying voices into young, adult, and senior groups.

Main Methods:

• Utilized artificial neural networks (ANN) and support vector machines (SVM) for classification.
• Extracted features from both the speech signal (MFCC) and the glottal signal (obtained via inverse filtering).
• Employed a wrapper approach for feature selection to identify the most relevant parameters.

Main Results:

• Classification was performed on three age groups: young (15-30), adult (31-60), and senior (61-90).
• Glottal signal features yielded the highest classification rate compared to MFCC alone or a combination of both.
• The proposed method demonstrates superior performance in vocal aging detection.

Conclusions:

• Features extracted from the glottal signal are highly effective for vocal aging classification.
• This approach offers a novel and accurate method for age-related voice analysis.
• The findings suggest a significant contribution to the field of speech processing and aging research.