The spectrum view lacks of precision in bass area.
IMO this is because the FFT / frequency points of the analysis should be placed using an exponential curve for the logarithmic view, so in the end, the stepping looks linear also in logarithmic mode, just like in linear mode.
Is this possible? If so, can you add such exp. curve?
An higher precision in bass would help a lot. Using the spectrum with 16384 window size, the precision currently is even uselessly high in the upper area, too.
I think it is technically possible, but you don’t encounter it very often, usually just a simple fft is used which will provide less resolution in regard to a logarithmic scale the lower the frequency is.
But for an fft you can’t just place the frequencies where you’d like - the “fft” is so fast compared to other fourier transforms because the frequencies are evenly spaced and have the window size as denomiator, other techniques are much slower. Also the response time / time window is another tradeoff point, you have to balance fast/slow response with resolution and vice versa. I think it should be possible to combine ffts of different resolutions (ala “bounded q transform”), so you have more resultion but slower response time in the bass freqs, and still could use the fast fft algo.
Ah, thank you for that explanation 
Did not know that.
Maybe it could be possible then to varying the window size for specific areas?
Like:
20-60 -> 64000
61-120 -> 32000
120-2000 -> 16000
2000-10000 -> 8000
10000- -> 4000
Oops, do you think that would be possible?
+1 for this.
Another suggestion for the spectrum view: