WHAT TO LISTEN FOR?
Optimal azimuth minimizes crosstalk and phase mis-alignment between channels. Therefore, optimal azimuth will cause sonic images to sound their tightest (laterally) and most coherent. Inaccurate azimuth causes sonic images to sound "fat", wider than in reality.
WHAT TYPE OF MUSIC?
Since we're listening for crosstalk/phase coherency between channels, we obviously need recordings with information from the same instrument/voice in both channels. This could be a mono recording or a stereo one with an instrument/vocalist placed in the center.
Since we're listening for right-sized images, small scale acoustic instruments and/or high quality vocalists work well because we have a natural sense of how "fat" they should sound. We all know how wide a real flute, violin or soprano sounds in real life (I hope). Nobody knows how wide an electric guitar or keyboard should sound... it depends on the number and placement of speakers and mikes and how the inputs were mixed. Leave the amplified/processed recordings on the shelf when fine tuning your rig. They're fun when you're in the mood for that sort of thing but they provide little useful information.
We're listening for directional and size information. Human ears take their directional and size cues from timing differentials between our L and R ears. Such differentials are more easily heard and measured with higher frequencies than lower frequencies. It's simple physics. You can't accurately measure a 4 inch differential with a wavelength that's 40 inches long. Therefore, go for a solo oboe or soprano, not a pipe organ, basso profundo or the Mormon Tabernacle Choir.
Once you get the hang of it you can adjust azimuth with many records but it's easier if you follow the above guidelines. They're based on the physics of the stylus/groove interaction and human hearing, confirmed by my experience setting azimuth both by ear and by electronic measurement. I can pretty much duplicate electronic measurement by listening, with far less time wasted.
