<div>Hi,</div>
<div> </div>
<div>Thanks for the information actually I am using wireless sensor data using mote. It is giving the gyroscope and accelerometer data. I am very new to DSP and I have to calculate the PSD power density spectral of 8, 10 and 20 samples of incomming data. While going through the tutorials and book I figured that i can either take the auto correlation of signals which is nothing but by taking fft and taking real and imaginary frquency components magnitude. Is there direct c routine which i can invoke from my application for maemo to calculate the PSD. I wanted it for the application running on N800 (developed using maemo) if there is such routine for ARM please let me know.</div>
<div> </div>
<div>Thanks and Regards,</div>
<div>Nisha<br><br> </div>
<div><span class="gmail_quote">On 4/23/08, <b class="gmail_sendername">David Huggins-Daines</b> <<a href="mailto:dhuggins@cs.cmu.edu">dhuggins@cs.cmu.edu</a>> wrote:</span>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">On Wed, 2008-04-23 at 14:15 -0500, nisha jain wrote:<br><br>> to port it to N800 later on...Please let me know if some one has<br>
> developed such<br>> application and used fft calculation for discrete signals...<br><br>If you are dealing with real-valued data (e.g. audio), you can use the<br>fixed-point FFT routine in CMU Sphinx if you like. It's not<br>
specifically optimized for the ARM but it's pretty fast. Browse the<br>code at:<br><br><a href="http://cmusphinx.svn.sourceforge.net/viewvc/cmusphinx/trunk/sphinxbase/src/libsphinxbase/fe/fe_sigproc.c?view=markup">http://cmusphinx.svn.sourceforge.net/viewvc/cmusphinx/trunk/sphinxbase/src/libsphinxbase/fe/fe_sigproc.c?view=markup</a><br>
<br>The functions you need are fe_create_twiddle() and fe_fft_real(). They<br>come in 16 and 32-bit versions, based on the FIXED16 preprocessor<br>symbol. The 32-bit version is a bit confusing since it uses 20.12<br>fixed-point format, so I'd recommend the 16-bit one (which is faster<br>
anyway).<br><br>You'll have to do a bit of work to separate it from the rest of the<br>front end processing, but that shouldn't be hard.<br><br>Better would be to use TI's dsplib on the DSP core, but I still haven't<br>
gotten around to figuring that out.<br><br></blockquote></div><br>