The off-axis dispersion characteristics of a loudspeaker will let you antecipate how it will perform in a room above ~300Hz because it is related to early reflections.I think there has to be a distinction between individual equipment measurements (test bench, anechoic, etc.) and that of a acoustic in room measurements of a whole system at listening position.
Where the former are next to useless above a certain acceptable level of performance to predict sound quality. The latter can describe a fair bit about the sound. However what we listen to is music, not sound, hence listening tests are vital.
Floyd Toole has done extensive research in these matters:
Toole has concluded that normal reflections in a typical small living room seem not to interfere with perception of the recorded space. He has also determined that early lateral reflections (<50ms) have a beneficial effect on intelligibility similar to raising the dialog level, and that the reflection pattern is more important than reverberation.
This has led to Toole’s recommendation that too many or too few reflections can be a problem. In particular, acoustic absorption, diffusion, and reflection must be broadband, ideally starting below 200Hz. He pointed out that the typical 1˝ or 2˝ sound panel most often affixed to walls works only at relatively high frequencies, and acts to effectively turn down the tweeter with no effect on the midrange or upper bass, thus unbalancing the sound.
Toole reminded us that many of the practices to deaden a room’s acoustics came from standards for broadcast and recording control rooms, where sound details must be heard more clearly, not for general listening rooms.
Harman testing has shown that sonic anomalies which are audible while the listener is moving around become much less obvious once the listening position is essentially fixed (the listener sat down!). Toole spoke of comb filtering, and has deduced that it is an artifact of measurement, not an audible problem in normal listening. We are less sensitive to room resonances in music than in broadband noise—our hearing is sensitive to the frequency response effects, not the time domain ringing, particularly in a reflective space, due to our “multiple looks” at the signal (except in the bass range, where both are factors). Audibility thresholds decrease with repetition and duration in reflective rooms.
Humans seem to be able to adapt to a room’s acoustics, up to a point. Thus, we should be able to predict real-room performance from a proper set of anechoic measurements, except for bass issues below about 300Hz. In Harman’s listening tests, low-frequency performance accounts for about 30% of a speaker’s rating.
Harman testing has shown that speakers with a fairly flat on-axis response and with a smooth and gentle off-axis level fade and roll-off are preferred by expert and amateur listeners. A speaker with an off-axis response that differs substantially from its on-axis response cannot be corrected with an equalizer. In-room measurements lower the accuracy of the data since they include a mix of direct and reflected sound. Since with most speakers the reflected sound has a different frequency response than the direct sound, an equalizer will cause more problems than it will cure.
(…)
Regarding bass, the Schroeder frequency is that threshold above which room resonances are considered so close together that they can be ignored with respect to frequency response anomalies. Toole has found that this works for large rooms, but the formula for its calculation yields too low a frequency for small rooms. In fact, Toole believes that calculation of room resonances is irrelevant in real home listening rooms, because not all modes have an equal effect or are heard equally, since there are multiple sources (speakers), and because all the formulas are based on a source in a corner with the measurement taken in the opposite corner (where there is never a listener). This has led to Harman’s recommendation to use four subwoofers in the center of each wall to provide the most uniform bass over the largest listening area in the widest range of small-room sizes; it reduces, and sometimes eliminates, the need for bass EQ.
‘Conventioneering’ by David J. Weinberg in Multimedia Manufacturer Magazine, March/April 2006
I disagree with your last sentence; music is sound(s).The latter can describe a fair bit about the sound. However what we listen to is music, not sound, hence listening tests are vital.
The Encyclopedia Britannica describes music as the "art concerned with combining vocal or instrumental sounds for beauty of form or emotional expression, usually according to cultural standards of rhythm, melody, and, in most Western music, harmony. Both the simple folk song and the complex electronic composition belong to the same activity, music. Both are humanly engineered; both are conceptual and auditory, and these factors have been present in music of all styles and in all periods of history, throughout the world."