Phomchick wrote: "The single most important factor in audio stereo imaging, is increasing direct radiation and decreasing reflected radiation, by moving the speakers away from the back wall, and away from the side walls."
I would agree that minimizing early reflections is AMONG the most important factors, but I don’t know what is THE most important factor.
Warning: The following is fairly technical, please skip it if you prefer not to read technical explanations.
Very early reflections (within about .68 milliseconds of the direct sound) are arguably the most detrimental to imaging, as they occur before the Precedence Effect kicks in. (Sound travels about 9 inches in .68 milliseconds). These are usually reflections (or diffractions) associated with the front baffle and sometimes associated with the drivers themselves. This brief time interval is within the arrival time difference for off-centerline sounds that arrive first at one ear and then at the other, from which we compute the angle that the sound comes from, which is why they tend to be especially detrimental to precise imaging. The longer the time interval (up to .68 milliseconds) before a sound arrives at the farther ear, the further off-centerline the sound seems to come from. So as we get closer to the .68 millisecond threshold, a correspondingly further-around-to-the-side false localization cue is generated. I believe this is why narrow speakers generally (there are many exceptions) image better than wide ones: Because the cabinet edge diffraction happens earlier within that .68 millisecond window.
Next we have the early reflections off of room boundaries, and the sidewall reflections are probably the most significant. They tend to "broaden" the image size, which many people find subjectively pleasing (according to Toole), but they can also degrade clarity (according to Geddes and Griesinger). These early sidewall reflections also tend to reduce the soundstage depth. Along with the (subjectively more benign) floor and ceiling bounces, the early sidewall reflections tend to superimpose a "small room signature" on the sound, which tends to mask the soundstage on the recording.
Ime relatively late-onset (after about 10 milliseconds) reflections are generally not detrimental to image localization, and can be beneficial by increasing the sense of envelopment and immersion within the soundscape on the recording. I like to intentionally increase the amount of late-onset reverberant energy for this reason: The ear/brain system judges the room size by the time interval between the first-arrival sound and the "center of gravity" of the reflections. By pushing that "center of gravity" further back in time, we reduce the amount of "small room signature" that is super-imposed atop the soundstage on the recording. Incidentally this is what acousticians try to do when designing the control room for a recording studio, so that the engineers can clearly hear the soundstage on the recording without the control room’s signature being super-imposed on top of it.
Imo, ime, ymmv, fwiw, etc.
Duke