On-axis frequency response cannot predict in-room measurements but does predict perceived sound.

Evidence

As indicated in Figure 2.6(d), sound power matters progressively less with rising frequency. This explains why the frontal hemisphere data, exemplified by the early-reflections curve, is a better match to the room curve. The sound power curve overestimates the low-frequency room curves and both it and the early-reflections curve underestimate the high frequencies. This is a reason to blend in at least high-frequency data from the listening window to improve the frontal weighting of the data. In no case is the on-axis frequency response by itself a useful indicator of how loudspeakers will measure in rooms—although it is critical to anticipating how loudspeakers sound in rooms. It bears repeating: two ears and a brain are not equivalent to an omnidirectional microphone and an analyzer. — Toole, Floyd E, Sound reproduction- the acoustics and…, 1321

Categorization

• QUESTION: Why does sound power matter progressively less with rising frequency (referenced from Figure 2.6(d))? Worth tracing the underlying psychoacoustic reasoning for ear-directional high-frequency hearing.