Optimal selection of mother wavelet for accurate infant cry classification

Area of Science:

• Signal Processing
• Biomedical Engineering
• Data Analysis

Background:

• Wavelet theory is a powerful tool in signal and image processing.
• The selection of an appropriate mother wavelet significantly impacts analysis outcomes.
• Identifying the best mother wavelet for infant cry classification remains an open research question.

Purpose of the Study:

• To compare the performance of various mother wavelets.
• To identify the most suitable mother wavelet for accurate infant cry classification.
• To optimize wavelet selection through parameterization for improved analytical findings.

Main Methods:

• Wavelet packet transform was used to decompose infant cry signals into five levels.
• Energy and entropy features were extracted from different sub-bands of the cry signals.
• Four supervised neural network architectures were employed to test the effectiveness of extracted features.

Main Results:

• The study evaluated mother wavelets based on similarity, regularity, and classification accuracy.
• The Finite Impulse Response (FIR) based approximation of the Meyer wavelet demonstrated superior performance.
• Optimal wavelet selection through parameterization yielded better analytical results compared to random selection.

Conclusions:

• The FIR approximation of the Meyer wavelet is identified as the optimal choice for infant cry classification.
• This research provides a systematic approach for selecting mother wavelets in signal processing applications.
• Accurate infant cry analysis can be achieved through careful selection and parameterization of mother wavelets.