So, I am wanting to detect a low frequency below 40 hertz. sigmund~ does not seem to detect frequencies below 40. I am at a loss as to why this might be.

To get around this I am using a simple granular patch that pitch shifts the audio up two octaves. Seems to work. But any thoughts on why this might be?

I don't know why, guess the wavelength is too long for the FFT to follow the whole circle? But bigger window-sizes don't work neither...

Downsampling improves tracking of lower frequencies:

[sigmund~] in a [pd sub] with
[block~ 64 1 0.5] or 0.25 or 0.125 or 0.0625 ...

Ah, I think that is making sense.

Looking up downsampling.

The downsampling is working.

both fft- and most other methods are not able to do pitchtracking in the subbass region.

@raynovich @katjav is worth reading on this subject. Good links to SNAC and others..... and you will need some filtering when downsampling........... https://www.katjaas.nl/helmholtz/helmholtz.html
David

Thanks all. @obineg and @whale-av.

If I just want to get the fundamental pitch, I thinking pitch shifting the audio before checking will work. Any thoughts on that? As opposed to downsampling. Is one more efficient than the other?

if the pitchshifting method involves windowing techniques it might end up beeing not much better.

@obineg Good point. Though, I think for "just" pitch detection it seems to be ok. I could be wrong, but as I am looking at the files it seems to work.

also try frequency shifting and compare if it works better.

pd has that "olli´s polyphase IIR" based hilbert transformer somewhere, which goes down to well under 20 Hz.