Reaktor Wavelet Project
A Dual-Tree Complex Wavelet Packet Transform
This upload currently contains examples of 2 variants of the conventional Discrete Wavelet Transform (DWT). Both are described below. Hopefully more to come soon.
Dual-Tree Complex DWT - splits an audio signal into several frequency bands. Each band contains twice as much spectral information as the band below it (IE the highest band will contain all the frequencies from 1/4/ to 1/2 of the sampling rate, the next band will contain frequencies from 1/8 to 1/4, and so on and so forth.
The Dual-Tree Complex DWT is best when you need a limited number of frequency bands, such as in a spectral delay, or in a multi-band compressor.
Dual-Tree Complex Packet DWT - allows you to further split the signals from the first ensemble into smaller frequency bands. Allows for many more frequency bands than the version without packets.
Both ensembles include example displays, as well as an FFT analyzer. The FFT is useful for testing purposes. If you want to make changes to the structure, you can use the FFT to make sure that the signal is being put back together again properly (sometimes the distortion is very subtle and hard to hear).
Each ensemble can easily be expanded to add more frequency bands. Be aware that more bands means more delay, sometimes substantially more.
Special thanks to Blokdak for assistance with math, theory and coefficients.
Have suggestions or something to add? Please join us in this thread - https://www.native-instruments.com/forum/threads/discrete-wavelet-transform-in-reaktor.270332/
A short series of tutorials is forthcoming on the subject at reaktortutorials.com