Dept. of Phys., Hunter College and Graduate Center of CUNY, New York, NY 10021
The frequency content of many natural and man-made signals changes drastically with time and standard Fourier analysis does not fully describe such highly nonstationary signals. Among such signals are speech, sonar and radar, optical images, and biological and geophysical signals. In the past 10 years there have been dramatic strides made in our ability to understand and process such signals. The basic idea is to develop a method that describes the intensity of a signal jointly in time and frequency. This would give the frequency content at each instant of time and hence describe how the spectrum is changing in time. A review is presented of the ideas and methods that have been developed to describe a time-varying spectrum and their application is illustrated with concrete examples from a wide variety of fields.