Well I may have had a breakthrough of some sort, and maybe at least have a workaround.
I was swapping some ICs around and accidentally powered the system on with U26 removed (ALS175) and while the system post'd more or less normally, the speaker tones were played without the brief silence between notes, and the final tone continued to play, sounding quite a bit like my ongoing beeping problem.
So, I took some more recordings of the speaker tone and realized that it's actually repeating the final note from the system bootup, which is pic_freq/415 or A flat.
This was pretty intriguing, so I went into sound.inc and started tinkering with the values. I altered the tone values and the repeating note invariably ended up being the very last note played during BIOS post.
I've now set the only note to be played to "0", which has disabled all boot tones but I do hear a very low click as the speaker is enabled or disabled during boot.
I'm doing some more extensive testing but I'm pretty excited about this so far.
Now, I'm not a programmer by any stretch of the imagination unless Visual Basic 5 counts, but as I look through the code I notice that the notes definition is stored into the SI register. I wonder if some lingering data there is causing my problem farther on, should that register be cleared after the notes are pllayed?
I also notice the statements for turning off the speaker are different in the "beep" and "sound" code sections in sound.inc; in "beep" it's "xor al,03h ; turn off the speaker" and in "sound" it's "and al,0FCh ; turn off the speaker"