From a user perspective, all that you really need to know is:
The more negative the curve value, the more “concave” the curve is
The more positive the curve value, the more “convex” the curve is
When curve = 0, the fade is linear.
In this respect it is similar to the “Curve” slider found on many DJ mixers, but with a much greater range of possible settings than found on most mixers.
What? There’s no “curve” slider thing at all. There’s no intermediate option between Triangle and Hann. Mid–fade Adjust, which is what I’m asking about, is from -100.0% to 100.0%. Triangle (called Fade Up/Down) and Hann (called S–Curve Up/Down) are two separate options in the Fade Type dropdown. You must be using some version newer than 2.2.2.
Sorry, but the most accurate explanation I can give of how it works, is the code, and I’ve given you links to that twice.
Other than that, please refer to the use documentation: https://manual.audacityteam.org/man/adjustable_fade.html The user documentation is less precise / detailed, but easier to understand.
On the subject of time scale/pitch shift, some sample–level procedures can cause an effect that pitch shifts some sounds. For example squaring all samples, for a centered sine wave, produces a quieter, double frequency, non–centered sine wave. However on a combination of 2 sine waves it distorts by adding another sine wave, so it isn’t a reliable pitch shift. It’s some parameter of the Even Harmonics distortion, I guess. Other procedures may similarly triple sine wave frequency, but also causing distortion on other sounds than single sine waves. That’s because they are just functions that happen to transform sine waves like this. True pitch shift algorithms shift the samples in time.