Main site  Forum  Search

Email contact  Register  Log in

Topic: FFT source if available

Author	Message
alyver00 Joined: Sep 25, 2007 Posts: 9	Posted: Jan 6, 2008 6:12 PM    Post subject: FFT source if available Hi Jarek, i see that from the 1.0.9 to the 1.1pre7 the FFT was improved and simplified, there is the possibility to have the source code of the various class FFT, FFTFloat and Windowing? thanks it will much appreciated. alyver
jarek Joined: Oct 22, 2007 Posts: 1075	Posted: Jan 6, 2008 7:41 PM    Post subject: Hi Alyver, yes, the FFT have been improved, but after 1.0.9 bioera has not been open source any more. Is there anything in particular you are curious about? Jarek
alyver00 Joined: Sep 25, 2007 Posts: 9	Posted: Jan 8, 2008 2:13 PM    Post subject: Hi jarek, i'm curious about you fft implementation, how do you convert an array of data ( the eeg signal) in a fourier trasformed array without limitation for input signal rate. because i always use algorithm that have an fft resolution depending on the sampling rate of the signal: if my signal have a sr of 44100 hz and i analyze and transform a signal vector of 1024 sample i have a poor frequency resolution of (44100/1024) ~ 43 Hz eatch frequency bins. Can you tell me how to fourier transform a signal without depending on the signal rate? Thanks
jarek Joined: Oct 22, 2007 Posts: 1075	Posted: Jan 8, 2008 5:04 PM    Post subject: Hi Alyver, FFT always depends on signal rate, based on rate you calculate its frequency coefficients. But you can take longer window, e.g. calculate FFT on 8192 samples and the resolution will be 8 times better then with 1024. Jarek
alyver00 Joined: Sep 25, 2007 Posts: 9	Posted: Jan 8, 2008 5:18 PM    Post subject: Ok, i understand but i think a longer window will take more time to compute that use a smaller window with smaller sampling rate.How much is the sampling rate and the trasformation window do you use on bioera?
jarek Joined: Oct 22, 2007 Posts: 1075	Posted: Jan 8, 2008 6:16 PM    Post subject: Yes, longer window will require more computation, but on PC that is usually not a problem. BioEra can easily compute FFT with 8192 or more samples. I did some designs with real time sound processing to tune my guitar and it worked great. Do not remember now exactly what sampling rate I used, it was probably 8kHz, that is usually the lowest sound recording rate on Windows.