A Psychoacoustic Feature Extraction and Spatio-Temporal Analysis Framework for Continuous Aircraft Noise Monitoring

Area of Science:

• Acoustics and Signal Processing
• Environmental Monitoring
• Aerospace Engineering

Background:

• Traditional aircraft noise monitoring relies on scalar energy indicators, which lack spectral-temporal detail and perceptual attributes.
• Existing methods struggle to differentiate noise characteristics under high-density operations.

Purpose of the Study:

• To develop and validate a sensor-based framework for continuous aircraft noise monitoring using psychoacoustic features.
• To enhance the characterization and interpretability of aircraft noise beyond conventional energy metrics.
• To analyze the perceptual structure of aircraft noise, considering factors like loudness, sharpness, and tonality.

Main Methods:

• Deployed an automatic noise monitoring system (ISO 20906) for synchronous acoustic and ADS-B trajectory data acquisition.
• Developed a spatiotemporal fusion algorithm to link noise events with flight paths, followed by IQR-based data cleaning.
• Extracted psychoacoustic features (loudness, sharpness, roughness, fluctuation strength, tonality) from cleaned data.

Main Results:

• Psychoacoustic features retain significant discriminative information lost in scalar energy indicators.
• Fluctuation strength and tonality showed 15-69 times higher sensitivity than energy-based metrics.
• UMAP analysis revealed clear separation between new-generation and legacy aircraft in the psychoacoustic feature space, unlike energy-based methods.

Conclusions:

• The proposed framework enables integrating psychoacoustic feature extraction into continuous sensor-based aircraft noise monitoring.
• This approach enhances the resolution and interpretability of aircraft noise measurements in complex environments.
• Psychoacoustic analysis reveals limitations of energy-centric monitoring and highlights decoupling between physical design and perceptual noise attributes.