WAV Dither Export

We are assured that Audacity adds a tiny dither signal to plain 16-bit Exports to keep the 32-Float to 16 bit errors from adding up.

How do you know that? I should be able to see that, right?

This post’s original question was which dither personalty is used. If shaped, which shape?

However, I begin to wonder if 3.4.2 adds anything.


The easiest way probably would be to export a sweep with and without dithering enabled in Audacity’s settings, and then compare the files in a hex editor

Audacity will apply dither by default, unless you have just imported a file and then exported it right away. In this case Audacity is smart now (thanks, Paul) and won’t apply the dithering.

This is a “clean” 3.4.2, so the default Dither On should be there, but I can check.

I generated Silence and then exported it as a WAV file. I can’t find any additional noise, even after much amplifying.

But to the original idea. There is a version of noise (shaped??) that’s all high pitch and low pitch with nothing in the middle. It does its job and it’s harder to hear.

Is there a version that “knows” what the show volume is? That would seem to be insanely complicated, but may make the dither insertion less obvious.


I found the preference and I don’t know that this helps any.

Screen Shot 2024-03-01 at 8.58.08 AM

I see the dither explanation in the “32-Float” forum post.


This topic was automatically closed after 30 days. New replies are no longer allowed.