Detecting deep voids in urban areas remains a challenging problem for geophysical methods, particularly with conventional active seismic sources that lack the low-frequency energy necessary for greater depth penetration. To address this, we explored an alternative approach using seismic noise generated by the engine of an idle vehicle to supply the needed low frequencies and improve the detection capabilities for voids at depths beyond 8 m. A key obstacle in processing this seismic noise is the presence of cross-correlation residuals, which occur when traditional linear stacking methods struggle to achieve stable averages, particularly with shorter-than-usual time duration data. These residuals can interfere with the identification of void signatures, related to backscattering and ringing, making void detection challenging in dynamic urban settings. We used a symmetric autoencoder (SymAE), an unsupervised machine learning architecture, to identify void signatures. By training neural networks to disentangle coherent subsurface information from vehicle seismic noise, SymAE effectively minimized the effects of cross-correlation residuals and enhanced the extraction of backscatters and ringing patterns. By applying this technique in a semi-controlled field experiment, we successfully detected two known voids at approximately 10 m depth. We further validated this approach using realistic synthetic data generated from a recorded vehicle signature. SymAE reduces data requirements while improving detection accuracy, showing the potential for rapid, cost-effective, and reliable subsurface void detection for urban environments.