[Andrew Haley]

writes:
On Jun 4, 7:12 pm, Steven Sullivan wrote:

As for truncation being subjectively *preferable* to dithering (and
let's assume it was *good* dithering) ...oy vey. Do you have a
reference for this?

See Keith Howard's "Contingent Dither" article in
Stereophile: http://www.stereophile.com/features/705dither/ .

I guess it isn't so very surprising that people might prefer simple
truncation to dithered truncation. I believe that truncation is
sometimes preferred in low-bit ADCs because it results in an audibly
lower noise floor. It's not inconceivable that some people might
prefer undithered quantization in high-bit conversion, especially if
they ride the volume control in quiet passages.

It would be interesting to perform an experiment to find out where the
sensitivity threshold is. Can you tell the difference with 12-bit
truncation? 14-bit? 16-bit? And so on.

Andrew.