FFT vs Gabor Transform

Question

I am quite new to signal processing, so I would need some help regarding a project.

I am interested in increasing the bass of a song from Spotify. At first I was thinking that applying a FFT to it to get the frequency content and then adding a Low-Pass Filter would be enough. But I started reading from here: https://en.wikipedia.org/wiki/Time%E2%80%93frequency_analysis_for_music_signals

I am not sure what would be the best approach. Do I need something like a Gabor Transform or the FFT is just enough? Why is one better than the other and when should each one be used?

Answer 1

An IIR filter as a cascade of second-order-sections (sos) would be a solution to consider for this application over filtering with the DFT. Please refer to the famous Audio-EQ Cookbook by our Robert Bristow-Johnson. The "lowShelf" filter would be used to boost bass. Robert is quite active on this site and can probably answer other questions you have as you proceed.

More details on shelving filters for audio applications can be found here.

The second-order-section is also referred to as a "Biquad Filter" which the short name for the 2nd order IIR biquadratic section; a common implementation approach for higher order IIR filters, done as a cascade of 2nd order sections.

This post also details how to decompose a higher order IIR transfer function into second-order-sections using Matlab's tf2sos. Similar commands are available in Octave and Python.