The Use of LPC and Wavelet Transform for Influenza Disease Modeling

Area of Science:

• Biomedical Engineering
• Speech Processing
• Computational Linguistics

Background:

• Influenza infection can alter physiological and pathological speech characteristics.
• Accurate detection of influenza through speech analysis remains a challenge.
• Existing speech analysis methods may not fully capture subtle pathological changes.

Purpose of the Study:

• To develop and validate a novel system for modeling pathological speech features associated with influenza.
• To investigate the effectiveness of combining Discrete Wavelet Transform (DWT) and Linear Prediction Coding (LPC) for influenza detection.
• To compare the proposed system's performance against existing speech recognition systems.

Main Methods:

• Utilized a real-world database of recorded speech samples.
• Applied a novel combination of Discrete Wavelet Transform (DWT) for feature extraction and Linear Prediction Coding (LPC) for speech modeling.
• Conducted three verification experiments: Normal/Influenza, Smokers/Influenza, and Normal/Smokers.
• Calculated various classification scores to evaluate system performance.

Main Results:

• The proposed system achieved very high classification scores, particularly for the Normal/Influenza verification system.
• Demonstrated superior performance compared to other published speech recognition systems.
• The combined DWT and LPC approach effectively models pathological speech changes due to influenza.

Conclusions:

• The novel speech analysis system effectively models pathological features of influenza.
• The proposed method, combining DWT and LPC, offers a superior approach for influenza detection via speech.
• This research provides a promising tool for non-invasive influenza monitoring and diagnosis.