When I view a wav file of pink noise with the FFT type set to "Real FFT" and "logarithmic" either checked or unchecked for the frequency axis, the pink noise declines at 3 dB per octave on the peak hold graph.

However, only if "logarithmic" is checked and I switch to an FFT type of "Complex . . .", then the pink noise is displayed as flat on the peak hold graph. This is how I want to see the pink noise, but I don't understand what is happening.

There is linear and logarithmic on how the data is displayed on the frequency axis itself and there is linear and logarithmic on how the FFT is accumulated. The Spectrum Lab setting is dealing with the former.

Spectrafoo software (http://www.metric-halo.com/beta/SpectraFoo4_Manual.pdf) provides the option to use "Pink Noise is Flat" or "White Noise is Flat" as the accumulation modes. These show either the sum of power or maximum power per the linked pdf.

Smaart software (https://www.rationalacoustics.com/files ... entals.pdf) allows one to view the data scaled on equal intervals per Hz or equal intervals per octave. Equal intervals per octave ends up with pink noise as flat.

What is the correct way to show pink noise as flat in Spectrum Lab?