Abstract: On the basis of human ear recognition, the hammer sound was collected for concrete-filled steel tube of the same size, different working conditions and types, and the sound characteristic value was extracted by simulating the principle of human ear recognition, so as to accurately identify whether there was a void defect in concrete-filled steel tube. By inserting voids into concrete-filled steel tube in advance, the signals were collected and hammer sound admittance was introduced. Compared with concrete-filled steel tube without voids, the mfcc10max characteristic value extracted from hammer sound admittance can well identify voids or cavity defects in concrete-filled steel tube. When the mfcc10max characteristic value was greater than −0.4 and the mean value was greater than −0.23, it can be preliminarily determined that there was a void or cavitation defect in the hammer location. while the mfcc10max characteristic value was less than −0.5 and the mean value was −0.72 in the hammer acoustic admittance of CFST without cavity. Finally, this technique and index were applied to the cavity detection of tower concrete structure, and its applicability was proved. This paper quantified the identification of holes in concrete-filled steel tube (CFST) with specific indexes and values, improved the identification accuracy, and provided a strong basis for the identification of cavity defects in the subsequent field of CFST structures.