The interaction between an intense ultrasonic field and a soft solid can generate bubbles that expand and collapse, known as acoustic cavitation. Understanding this phenomenon is crucial in controlling the formation of bubbles in specific brain areas during transcranial ultrasound therapies. To achieve this control, establishing an acoustic intensity threshold, be-yond which cavitation is highly probable, becomes essential. As cavitation behavior can vary under identical experimental conditions, considering the probability of its occurrence becomes a crucial variable. This study introduces a passive detection system designed to identify acoustic cavitation, presenting the results of its implementation for a probabilistic analysis of cavitation phenomena. The setup comprises a high-power flat transducer operating at a frequency of 0.94 MHz, generating an acoustic field that traverses an agar-agar phantom. A secondary transducer, purpose-built for cavitation detection, captures the acoustic wave emitted by the phantom. The detection method involves analyzing the wave spectrum to identify the specific acoustic signature of bubbles: a subharmonic spectral component precisely at half the operating frequency. By conducting multiple iterations of the experiment, we determine how often cavitation is detected, thereby empirically establishing the likelihood of this phenomenon occurring. The results illustrate the correlation between the likelihood of cavitation occurrence and the maximum intensity of the applied acoustic field on the phantom. To elucidate the relationship between these variables, we introduce a model derived from calculating the effective volume where the acoustic field exceeds a threshold intensity value. This model aptly describes the experimental outcomes. Future work will extend this analysis to a transcranial HIFU experiment.