I've been playing around with one of these devices too (WTV020-SD-mini) and find it requires a bit of reverse engineering to figure out how it works. It doesn't pay to follow the data sheets out there too closely. As you've noted, the "BUSY" signal appears to be active high rather than low as the data sheet would seem to imply. Also, it's not really a BUSY signal but rather just a bit that says when the device is out of "standby" (low-power drain) mode. When a file is played, the BUSY line stays high for about one second after the play stops:
The top trace is the voltage on one of the speaker pins, the bottom trace is BUSY. The time scale is 1 second per division. The line actually goes low when the device switches to standby. BTW, it draws about 2 ma in standby as opposed to maybe 100 ma average when playing a loud sound.
So when trying to loop a sound using BUSY as a trigger there will be at least a one second gap in between plays. But there's another quirk--if you try to send another play command immediately after BUSY going low, it will be ignored. I find by trial and error that you need to delay about 10 ms after BUSY goes low to successfully wake the chip up again with a play command.
So to summarize, to loop a sound:
-- send play command
-- wait for BUSY to go low
-- wait 10 ms
-- send play command, etc.