Summary

Rick Lyons challenges a common statement that using the discrete Fourier transform (DFT) as a filter inherently frequency-translates an input spectrum to DC. The paper explains the correct interpretation of the DFT as a spectral sampler, clarifies the distinction between analysis and modulation, and describes practical implications for DFT-based filtering (including circular vs. linear convolution).

Key Takeaways

• Distinguish the DFT's role as a sampled-spectrum analyzer from an operation that frequency-translates signals to DC.
• Explain how DFT-based filtering corresponds to circular convolution and when to use overlap-add/overlap-save to achieve linear convolution.
• Demonstrate how windowing, zero-padding, and bin-centering affect spectral interpretation and apparent frequency shifts.
• Provide practical guidelines for implementing FFT-domain filters without misinterpreting spectrum location or introducing processing artifacts.

Who Should Read This

DSP engineers, graduate students, and practitioners implementing FFT/DFT-based filters in communications, radar, audio/speech, or research who want a precise, practical clarification of DFT behavior.

TimelessAdvanced

Topics