The self-oscillation of the cavitating vortices is one of the dangerous phenomena of hydraulic turbine operation near full-load conditions. This work is an attempt to generalize data and expand insight on the phenomenon of self-excited oscillations by comparing the experimental results obtained on a simplified turbine and scaled-down pump–turbine models. In both cases, a series of high-speed imaging was carried out, which made it possible to study these phenomena with high temporal resolution. The high-speed imaging data was subjected to additional processing such as binarization, cropping, and scaling. For a simplified turbine model, the volume of the vapor cavity was calculated based on the assumption of the axial symmetry of the cavity, after which fast Fourier transform (FFT) analysis was carried out. A proper orthogonal decomposition (POD) analysis was also performed to examine individual modes in the original digital imaging data. For the pump–turbine, visualization data on the cavitation cavity oscillations were supplemented by pressure measurements in the draft tube cone to determine the frequency characteristics. Based on obtained experimental data, an improved one-dimensional model describing the oscillations of the cavitation cavity arising behind the hydraulic turbine runner is proposed.