This extends from PRs #16 and #17 and should be evaluated after those are merged.
This PR adds a Continuous Wavelet Transform module. pywavelets is used to construct a set of filter kernels, which are then passed to filterbank. Note that the generated kernels will have very different lengths and delays according to their scale, so there is a min_phase option that will take the 'pristine' wavelet kernels and recalculate them as their min-phase equivalents. min-phase kernels have worse performance (weaker stop-band, more phase distortion in pass-band) but have much less delay so the synchronization between e.g. alpha decrease and gamma increase is preserved.
This extends from PRs #16 and #17 and should be evaluated after those are merged.
This PR adds a Continuous Wavelet Transform module.
pywaveletsis used to construct a set of filter kernels, which are then passed tofilterbank. Note that the generated kernels will have very different lengths and delays according to their scale, so there is amin_phaseoption that will take the 'pristine' wavelet kernels and recalculate them as their min-phase equivalents. min-phase kernels have worse performance (weaker stop-band, more phase distortion in pass-band) but have much less delay so the synchronization between e.g. alpha decrease and gamma increase is preserved.