Many years ago, a magazine did a review of a Dunlavy speaker with a remarkably flat response, both on and off axis, very little resonance evident in the waterfall plot, and the closest thing I ever saw to the ideal impulse response. I thought it showed promise of being the best speaker I had heard until I actually heard it; it wasn’t bad, but it sounded dry, and lifeless compared to some speakers that did not test so well.
We should also keep in mind that frequency response tests are only rough approximations given that ideal testing setups are hard to achieve. In Washington State, there is a testing company that has a test chamber inside of a nuclear reactor that was never commissioned. The chamber is something like 500 feet long and nearly as wide. The company guarantees accuracy of a speaker response test down to something like 35 hz. This makes me wonder about tests done under less ideal conditions.y
