In-Car Sound Signal Reconstruction Based on Order Extraction and Overlap-Add Algorithm

Abstract:

To enhance the reconstruction accuracy of in-vehicle acoustic signals in automotive NVH performance development， this study presents a method based on order extraction and overlap-add. The method extracts rotational speed curves corresponding to characteristic orders in time-frequency representations by combining path tracking with short-time Fourier transform， considering the order characteristics of rotating machinery sound sources. Signal reconstruction is accomplished through waveform similarity matching within a specific rotational speed range and an overlap-add algorithm. Furthermore， psychoacoustic evaluation metrics quantify perceptual similarity using objective parameters. The proposed method demonstrates superior performance in reconstruction accuracy and perceptual consistency， establishing a novel technical approach for in-car sound signal simulation.