Not a math expert (or an software engineering expert, tho I am an Engineering student). But I will try to explain this in my own way.
To sum up what jakko said, “peak” in equalization that is born to solve the “dip” in the headphone response is bad since it introduces unwanted ringing effects. The new algorithm was made so that the equalized response will avoid this to happen. Here is a breakdown:
- Smooth out the equalized response using normal means.
- Scan the frequency response from left to right and from right to left to find the “slope” and beat it down.
2.1. From my assumption, the reason why you have to do this is that you don’t have a real way to detect “peak” and “dips” when it comes to programming. In other words, you have to come up with a way to do this.
In jakko case here, he scan the smoothed equalized response from 0Hz to 10000Hz and vice versa, doing the opposite. This is due to the fact that a “peak” occurs if both side of such “peak” has a steep slope.
2.2. You would notice that there is the green zone that he dubbed as “safe zone”. This just means that after scanning left and right, his algorithm will add those zone so that AutoEQ knows that it should not include these in the limiting process which happens right after. - Limit equalizing those peak, as it is what you want to avoid
Again, note that this is based on my assumption on how the algorithm works, and might not be correct.