Fourier analysis of continuous data using DFTs

Traditionally the discrete Fourier transform (DFT) has been viewed as a mathematical transform. L.R. Rabiner and B. Gold (1975) have pointed out that, for continuous data, the DFT can be viewed as a bank of digital filters. The author looks at the issues related to this view of the DFT. Specifically, he reviews the resampling process associated with performing the DFT on an infinitely long stream of data and points out that the key issue is aliasing. He also reviews an efficient means for achieving locally very high resolution spectrum analysis while the complete spectrum is monitored. This is the concept of post-DFT zoom.<
>