My brother (MIBRO) loaded a 29-second clip of Toccata and Fugue in D Minor onto the little sound player. He was pleased with the playback quality, but we both knew it the sound would have to be amplified for sufficient spookiness.

I clipped off the little speaker and tried to measure the max levels on an oscilloscope. My goal was not to exceed 0.3V. This was difficult, as the dynamic range was quite wide. Eventually I just soldered a chunk of AUX cord with 1/8 in audio plug on the end to where the speakers had connected. It was a stereo plug, so I soldered both channels to one side and return to the other.

I then plugged this in to an old iPhone dock with nice speakers that MIBRO had provided. I turned up the volume on the little sound unit as high as I dared without noting any distortion. I then hot-glued the volume in place to prevent damage and distortion.

When I hit the button, the Sound Unit played the music. Holding the button down didn't cause any trouble, so substituting a relay for the switch would be straightforward-- that is, if the relay was just activated by the skeleton power supply going on, it wouldn't need a timer to switch it off before the music ended. 

The little sound unit would charge just fine, but when I attempted to program another sound file into it (by hooking up the provided USB cable and turning it ON), the output would go to some DC voltage (like 2 or 3), the computer would not recognize the device, and the amplifier chip got hot. I don't understand how cutting off the speaker would cause this, but I am pretty sure that I wasn't shorting the output-- it was hooked to my O-scope. 

Post Script: OHHHH. The O-scope was a computer one--- there was a hidden short when I connected the device to the computer and the computer oscilloscope to the device. I had encountered this once before. Observation affects the observed.