Crossfeed, loudness, and the things headphones get wrong
Two DSPi features that exist because human ears are not line-level inputs.
Two of the DSPi features people ask about least and benefit from most are headphone crossfeed and loudness compensation. Both exist for the same reason: a recording is mixed for one listening situation and you are almost never in that situation.
Crossfeed
On speakers, your left ear hears the right speaker — slightly later, slightly duller, slightly quieter. Headphones remove that entirely, which is why a hard-panned 1960s mix can feel like it is happening inside your skull rather than in front of you.
DSPi’s crossfeed is BS2B-based, with interaural time delay, so the bleed arrives late the way it would in a room rather than simultaneously. Three classic presets cover most listening, and every parameter is exposed if you want to build your own.
Loudness
Human hearing is not flat, and it is differently not-flat at different levels. Bass and treble fall away faster than the midrange as you turn things down, which is why quiet listening sounds thin.
DSPi implements compensation against the ISO 226:2003 equal-loudness contours, tied to the volume control, so the correction tracks how loud you are actually listening rather than applying a fixed smile curve. Turn it up and the correction backs off on its own.
Neither of these is a tone control. They are corrections for a known, measurable mismatch between how the recording was made and how you are hearing it.