The laser-induced cavitation bubble collapse noise spectrum in the liquid are studied experimentally by using a piezoelectric transducer (PTZ) hydrophone. Experimental methods and a cavitation detection system are designed to investigate bubble collapse noise. The wavelet transformation, fast Fourier transform (FFT) and other signal analysis methods are used to monitor the cavitation noise and reduce wavelet noise, analyze power spectrum. The result shows that application of wavelet denoising in cavitation bubble collapse noise spectrum analysis is an effective method. In addition, the designed software can quickly analyze the spectral characteristics of bubble noise signal, and it can detect and monitor cavitation on-line. The results have certain significance on the warning and researching of turbine cavitation, propeller cavitation and other water conservancy machineries.