Possimpable-
> I want to take a sound file, and geometrically flip
every frequency
> around, say, A (440Hz). So A# becomes Ab, B becomes G, C becomes
> Gb... To basically flip the frequency scale (f' = 440/f). It's a
> similar concept to voice-inversion-scrambling. The only way I know
> how to approximate this effect is by producing a spectrogram (with
> a logarithmic frequency axis), vertically flipping the spectrogram,
> and resynthesizing it into sound with something like ARSS or
> AudioPaint. Is there a better way? And if not, what program can I
> use to generate high resolution spectrograms (large image files),
> so I can produce better results with my current process?
The following difference equation will "spectrally invert" a signal:
y[n] = x1[n]*(-1)^n
So that's a time domain method, as opposed to what you describe. I have
posted a
plot here:
http://www.signalogic.com/images/spectral_inversion.jpg
showing a speech signal processed this way. You mentioned a log freq axis, I
don't
know why that would matter -- it's just a function of the display type that
you use.
-Jeff
Re: How do I "flip" the frequencies of a sound file? What I mean is...
Started by ●December 16, 2009