The invention discloses an acoustic method for identifying shape of an object in real time. The method comprises the following steps: 1, obtaining a training set and a test set of a neural network; 2,inputting the training set into an acoustic super neural network, and forming certain sound pressure distribution on a detection surface after the training set passes through a multi-layer acoustic super surface; 3, dividing the detection surface into N regions, and obtaining a total sound energy value of the N regions; 4, calculating an error between the obtained N values and a label of the object, and calculating a gradient of phase modulation applied by each super neuron by using the error, thereby updating the gradient, updating the phase until stable output is obtained, and correctly identifying a target object; 5, determining a phase offset value of each unit on the super surface through the updated phase value, and making the acoustic super surface according to the phase offset value; and 6, placing the acoustic super surface at a proper position, and after the sound wave passes through the acoustic super surface, identifying the target object by a probe positioned behind the acoustic super surface.