[sdiy] Harmonic bandwidth

[Richard Wentk]

On 7 Jan 2008, at 18:08, Ben Lincoln wrote:

> Yes, sorry, I should have phrased what I wrote in a better way. I  
> agree that the main part of his approach is to replace pure  
> harmonics with clusters of closely-spaced (frequency-wise) sine waves.

The way to do this isn't to resynthesise using a sine bank, but to  
resynthesise using noise passed through a bank of very high Q  
filters. Depending on the noise spectrum and the filter Q you can add  
different kinds of 'fat'.

> What I meant was that since he lists the phase aspect as a separate  
> step, that he was using it to "smear" the harmonics in a different  
> way to alter or increase the complexity of the effect that he  
> obtains with the previous step.

Slightly randomising either frequency or phase components in an IFFT  
creates a chorus effect, providing it's done for the entire duration  
of a sample and not just for a single cycle.

A single cycle wavetable oscillator cannot reproduce this kind of  
fat, by definition, because it's a dynamic and not a static effect.

The chorus/fat effect is created by constantly varying changes in  
phase and frequency, not the static distribution of the overtones.

Richard