During the vaccine production through the chick embryo cultivation method, harmful cracks may occur from the perforation of a trocar on the eggshell, around the impact hole, leading to the failure of cultivation. To detect the perforative cracks, this study proposes a method based on acoustic responses. By stimulating the embryo eggs and collecting the acoustic signals, 7 characteristic values were extracted from the time and the frequency domains: The maximum value in the time domain; The difference in the time domain; The frequency-domain peaks, 870 Hz, 1250 Hz, 1470 Hz and 1770 Hz; The mean value of the waveform. Statistical difference analysis found that all the above-mentioned characteristic values revealed differences. Then, the support vector machine and confusion matrix were combined, and the differentiation was achieved. In the statistical difference analysis, the p values of the characteristic values were all less than 0.01, indicating differences. The demarcation lines of all characteristic values obtained via binary classification through the support vector machine were as follows: TM:0.4114, TD:0.7528, FP870:5.716, FP1250:1.18, FP1470:2.96, FP1770:9.206, WM:0.6564. The weighted composite scores of the characteristic values obtained through the confusion matrix were as follows: TM: 99.18 %, TD: 99.16 %, FP870:90.56 %, FP1250:96.73 %, FP1470:98.32 %, FP1770:92.56 %, WM: 99.16 %. Experimental results demonstrated that this study could effectively distinguish qualified perforated embryo eggs from cracked ones, with the experimental prototype possessing the potential to be upgraded into an online detecting system, which comes with great significance in theory and practice for the development of the vaccine cultivation industry.
