Last time I was working on sound visualization, after testing with real-life data (yes, music 😉) and testing out various visualization shaders, I came to a conclusion that I approached it from a too scientific point of view.

The result was fully proper spectrogram - but not so useful for visualization purposes.

So, now I've returned to it - but this time focusing on achieving more visually appealing results,

easier to read by a human. I wanted to make it similar to what Inigo Quilez is doing in his ShaderToy , but I couldn't find an exact way he is treating the data, so I had to come up with my own approach.

One thing still applies: the best way is to use a frequencies analysis, through FFT.…