Re: [fluid-dev] Purpose of dither?

[Z F]

--- Mihail Zenkov <address@hidden> wrote:

> On Wed, 9 May 2007 01:36:13 +0100
> Miguel Lobo <address@hidden> wrote:
> 
> > > Yes, you absolutely right. But look at that code again:
> > >
> > >   for (c = 0; c < DITHER_CHANNELS; c++) {
> > >     dp = 0;
> > >     for (i = 0; i < DITHER_SIZE-1; i++) {
> > >       d = rand() / (float)RAND_MAX - 0.5f;
> > >       rand_table[c][i] = d - dp;
> > >       dp = d;
> > >     }
> > >     rand_table[c][DITHER_SIZE-1] = 0 - dp;
> > >
> > > We subtract previous dither value from current to prevent noise
> > > modulation. I use dither table to get maximum speed.
> > 
> > I actually felt curious about the subtraction of the previous value
> too.  What 
> > do you mean by "noise modulation"?  (And yes, this time I have
> looked it up 
> > in Wikipedia ;)).
> 
> Roughly, dither convert harmonic distortion to noise. If we just add
> noise (random) to signal, distortion gone but noise will by modulated
> by signal (correlate with signal). This noise not fully random. This
> is
> rectangle dithering.
> 
> To prevent this modulation and get pure white (fully random) noise
> (it
> much better for human ears) not correlated with signal, we consider
> previous dither value.

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


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com