If background sound starts at the MPU power on PIA test and never stops, its usually a connector problem. The sound board (can) get stimulated during the power on test and then should be shut up when attract mode hits. If it does not, I'd focus on connectors first. If your solenoids are all working properly, then check the male headers on the sound board.

C10 is a tantalum bypass on the 12v supply. Not likely to be the cause of a problem, but should be fixed. Instead of using a tantalum cap, install a 10uF 25v+ electrolytic cap. It can deal with the ripple and not explode in a fireball like the tantalum ones do.

On the schematic drawing in the bottom left corner you will see the inputs from the computer board. They are labeled "SOUND-SEL" When the game is in sound test or solenoid test with a logic probe check to make sure all these sound selects are changing states. You can go right to the 6803 chip and check there.

NOTICE how two of the sound select pins sneak off into the reset section. I am not sure exactly what Bally is doing there, perhaps waiting for the PIA power on test to reset the board fully? Anyways if Q6 is shorted it would force a sound select HIGH and make the sound board play wrong sounds.