Hi, I'm giving an hand for a project that intend to transform
fingerprints to an image processed by animating the points found by
libfprint with the audio analysis of the print's owner voice.
Each point animation is rotating by a speed defined with amplitude of
the frequency where near the center is the lowest freq and the perimeter
is the highest.
That part was relatively easy to make, it's just a matter of computing
the sum of all magnitudes for each frequency range, with using fft,
filling a table with all magnitudes and summing all the values, and
after, making an interpolation with the current magnitude, at each
block, giving the new z angular reference to a Gem object that is
rendering on a simple buffered gemwin and capture an image.
Sampling rate have to be set to 48khz to get an integer number of
blocks, for an image rate of 24i/s... (48000/24=2000 samples)
Now I'd like to get the frequency range of the voice signal, before
making the magnitude analysis, to not get 'dead frequencies' in the
animation.
Also the point's radius will have to change following the current
deviation from the mean frequency for the analyzed range.
I've attached an archive with patches I've made for processing the
animation.
So my question is in the title, how could I get the mean frequency of
the whole signal, and from each frequency range?
That question won't come alone, I'm using Gem for processing the image
because I know it well enough, but the rendering isn't really amazing.
Each image are written in the hard drive with [pix_write] with [file
img/myimage 100(, and I'm never sure about how pix_write will find the
img/ folder that is into the same folder than the main patch, and the
best quality of a rendered jpg image is really dirty.
I'd like to use tiff images instead but I haven't found a CLI yet for
building the video with, mencoder can only handle the jpeg from pix_write.
Also the [pix_snap]--[pix_record] solution is not stable and
compatible enough for this installation.
Sorry to be out of topic but anyone knows a program I could handle
in a script that builds videos from tiff images?
And finally, does anyone know another solution for procesing the
image, with other pd externals, or external programs we can handle from pd?
Thanks.