In the paper Calculation of a constant Q spectral transform - J.C.Brown it is mentioned

The conventional linear frequency representation given by the discrete Fourier transform gives rise to a constant separation between components for musical sounds consisting of harmonic components

Since DFT requires the samples to be equally separated how would we always get constant separation between the frequencies ?


My guess is that Ms. Brown is referring to the notion shown in the figure below. The top panel shows the spectral magnitude of a sawtooth sequence, that repeats five times per second (fundamental freq = 5 Hz), with a linear DFT freq axis. Notice the "constant separation between harmonic components." The figure's bottom panel shows the same spec magnitude with a logarithmic freq axis. There the frequency separation between harmonic components is not constant.

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.