Re: [fluid-dev] Purpose of dither?

[Josh Green]

On Wed, 2007-05-09 at 05:56 -0700, Z F wrote:
> This is exactly my point. But for this to be correct two calls to
> a random generator per table entry is needed. If you do only one call
> and use a the second random number from the previous table entry you
> introduce correlations into the noise which means that the noise is not
> 
> white. Now, it could be a design feature or a mistake I do not know.
> >From what I understand about dithering, it is a mistake, not design and
> yes it increases the speed of a factor of two, but this is, probably,
> not what you wanted.
> 
> I addition, I do not know if this produces any noticable change at all,
> but since I am against dithering, I could be biased on that....
> 
> ZF
> 

I appreciate the discussion on the dithering additions made by Mihail.
I'm no DSP expert unfortunately, so its nice to have others around who
know more than me and are interested in FluidSynth.  I can conceptually
understand how converting from a higher bit width to a lower could
introduce distortions and how these distortions could contain periodic
patterns and thus heard as new frequencies in the output.  It also makes
sense that its better to have a small amount of white noise distortion
over a small amount of periodic distortion.  For this reason and the
fact that it sounds like other software generally follows this practice,
I think it should be the default in FluidSynth, provided the
implementation is correct.  Of note, is that when using Jack output, the
dithering is not done by FluidSynth (but likely by Jack).

Perhaps it would put your mind at ease if there was a compile time
switch to disable dithering?

Best regards,
        Josh Green