A signal processing technique that decomposes complex, dynamically changing non-stationary signals from machine components such as bearings into different scales by means of a continuous wavelet transform. The envelope signal in each scale is then calculated from the modulus of the wavelet coefficients. Subsequently, Fourier transform is performed repetitively on the envelope of the signal at each scale, resulting in an "envelope spectrum" of the original signal at the various scales. The final output is a three-dimensional scale-frequency map that indicates the intensity and location of the defect-related frequency lines. The technique is generic in nature, and applicable not only to machine condition monitoring, but also to the health monitoring of a wide range of dynamic systems, including human beings.