the wavelet stuff has really fascinated me, and i feel i’ve started on a long and interesting journey!
they are a bit hard to describe, and i’m not really a good explainer. it’s much easier to get the ‘feel of them’ when doing hands-on experimentation, and seeing/hearing results there and then…
so i updated the haar wavelet analyzer
cyan: incoming audio signal (in this case me playing around with sine waves with reaper’s reasynth)
white: the wavelet coefficients, ‘levels’ (or scales) from top to bottom (higher frequencies), wavelet magnitude displayed by rectangle intensity.
yellow: the same wavelet coefficients, unmodified from the analysis buffer, left to right, y = level/magnitude.
and a reaper test project.
what i’m looking at now, is different ways of (ab-)using this thing.
first, i think this can be used for some interesting analysis of the audio/signal. there is both timing and frequency info in there. and i have been thinking about searching, and what can be done with that (pitch detection?)
and, some interesting effects can be made, by modifying the coefficients, and then running them through the inverse transform, to get ‘normal’ audio.
then, maybe we can just skip the initial transform, fill the analysis with our own specific coefficients, and transform ‘back’ to audio, to make a weird kind of synthesizer. an additive synthesis variant.
finally, we can transform multiple audio streams, combine the coefficients in various ways, morphing, crossfading different parts of the spectrum, comparing, ..
lots of ideas…
will be posting new discoveries when/if i find them, and much more of these jesusonic effects!