There has been no effective method to monitor the slope instability and collapse of the muddy submarine channels so far. To predict and warn the instability and collapse of muddy submarine channel slope，the four features extracted by gray level co-occurrence matrix are used as recognition features，and the distance similarity is used for voiceprint feature matching，a voiceprint recognition system is designed and completed for the detection of instability and slumping of channel slopes. After inputting the sub-bottom profile images of the side slope of the channel to be detected into the system，it can quickly locate and display the image similar to the slope instability and collapse image. The test results show that the system has a high recognition rate，which is helpful for channel management and engineering technical personnel to detect the stability of channel slopes.