In this paper, an automatic fishing machine based on acoustic conditioning was developed and tested in a reservoir. The floating catcher is remotely and automatically controlled to operate underwater speakers, feeders and underwater cameras that monitor fish behavior in real time. The open net fence installed underneath the system can be automatically closed to catch captured fish. And in this paper, we monitor and discuss the dynamics of fish populations in the net box. With classic conditional reflexes, fish can associate sound signals with food. Also, in order to strengthen the conditioned behavior of fish swimming freely in the reservoir, we employ regular training. With the utility of an audible signal which is to attract fish around the catcher, the catcher can catch fish and automatically close the net fence around them: fish catches in this particular event are 2.5 times as much as ones in a controlled event. In controlled events, the net fence is automatically closed without summoning fish.