How would I use this FFT for my 20khz sampled audio signal? about dblclockfft HOT 3 CLOSED

Sounds good. Let me know how it works out for you.

Some insights you might wish to consider:

• The '-k 5000' is really pretty irrelevant. The max value supported here is '-k 3'. Larger numbers won't hurt though, so ... you are fine as you have it.
• When using '-m 6', you'll be throwing most of your PMIC bits away. I'd be tempted to use '-m 12' and then only throw bits away at the end of the operation.
• If your goal is to detect the presence of a tone, you might wish to look into windowing, and/or oversampling.

Dan

from dblclockfft.

1. Ahh that makes more sense... Hahaha i was wondering why my output did not change
2. Sounds good, i will probably use 12 bits, find the abs and then bitshift it down back to 6 bits
3. Thanks for the heads up, for the moment i am crudely hardcoding notes into my program to detect, however i might need to use cents over a pitch method eventually, may i clarify that oversampling is just feeding my fft more points than my source?
   For e.g. clocking my fft at 40Khz when my source is 20Khz

But overall, wow what an amazing tool you have built... Really saving my life here thanks!

from dblclockfft.

You might find this article valuable. It explains some of the issues associated with overlap and oversampling. You might also find rudimentary logarithms to be valuable when trying to work with large dynamic ranges.

Good luck!

Dan

from dblclockfft.