An Optimized Hybrid Approach for Reducing Computational Overheads and Evaluating Audio Signal Characteristics in Wireless Acoustic Sensor Networks

Abstract

This paper presents a hybrid system designed to analyze multiple properties of audio signals while minimizing quality losses during transmission over Wireless Acoustic Sensor Networks (WASNs). The proposed system operates in two phases. In the first phase, audio signal quality is evaluated using key parameters such as packet loss ratio, signal-to-noise ratio (SNR), peak signal-to-noise ratio (PSNR) and signal fidelity. The experimental results of proposed method reveal an increased packet loss ratio, reduced PSNR and lower signal fidelity degraded audio quality. An acceptable threshold is established to maintain quality, though network traffic exceeding this threshold negatively impacts performance. To address this, the system incorporates controls for packet loss, SNR, PSNR and fidelity, ensuring the transmitted audio maintains parity with the source. A WASN framework is introduced for distributed and efficient audio property analysis in the second phase. The framework employs feature extraction techniques, including Mel Frequency Cepstral Coefficient (MFCC) and Power Normalized Cepstral Coefficient (PNCC), alongside other existing methods, to extract comprehensive features from audio signals. Combining quality assessment and distributed analysis, this hybrid system provides a robust solution for enhancing audio signal processing within dynamic and resource-constrained network environments.