Hi to all, i'm starting to use pd and i think it's a very good program. I have a little problem: i have some audio signal from 0.2hz to 100 Hz in my input, i need analyze it and takeout the dominant frequency with the highest amplitude, but i need to do this with a precision of 1hz or less is better.
Who can i do? i can use the rfft but i don't have this precision.
Thanks

try [fiddle~]

not sure how you're going to remove the frequency though. you mean, you need to remove, for example 76hz, but leave 75 and 77hz intact? i'd love to be able to do that, but still never found a way.

are you needing to be doing this in realtime?

i have an EEG signal from 0.2 to 100hz i need something to find the dominant frequency with the highest amplitude and visualize in a number box the frequency, something like a spectrogram but with a numeric output of the most power frequency. I need to do this in realtime but few ms of latency for now is not important.
My project is to do a biofeedback, i build the soundcardEEG from the OpenEEG project, so i have an eeg signal modulated in AM. I do the demodulation with a Hilbert transformation, with some lowpass and bandpass. Now i have to drive binaural beats with the dominant frequency "came out of my head", the realtime is only important for other analysis like the p300 potential, for see only the rhythm of the brain i don't need a "real"realtime, 1 second of latency isn't a problem.

i will try what is fiddle~ thanks

fiddle~ i think has a standard blocksize of 512, which means about 1.2ms latency on 44.1khz soundcards.

i'm starting using fiddle~ but i can't understand one thing: the help file say: The minimum frequency that Fiddle will report is 2-1/2 cycles per analysis windows, or about 108hz. i can't understand what does it mean for report the "minimum frequency". with fiddle i also can change the blocksize until 2048.

That's fixed in [fiddle!] unfortunately.

For such low frequencies you might need to look at other frequency detection methods.

Period detection with integrating peak squared after zero crossing
Phase lock/tracking loop
Autocorrelation
FFT with a very big block size
A narrow comb/filter bank method

start reading here

http://cnx.org/content/m11714/latest/
http://www.cs.uregina.ca/Research/Techreports/2003-06.pdf