Several methods have been developed for simulating non-stationary and non-Gaussian processes in packaging vibration testing, encompassing unique methods for the segmentation of road vehicle vibrations. However, only a limited number of those consider spectral characteristics. Thus, the current paper introduces a novel segmentation algorithm conducted in the time-frequency domain. The spectral characteristics obtained by short-time Fourier transform are compared by multiple hypothesis tests to find changepoints in a wheeled vehicle vibration sample. Different post hoc procedures are introduced against the inflating Type I. error.