Cavitation is a common but undesirable phenomenon in the operation of centrifugal pumps. Methods to detect cavitation are needed to ensure safe operation. This study contributes to the approach of using acoustic signals for cavitation detection. Mainly focusing on microphone measurements, the effect of cavitation on frequency spectra at different positions around an industrial centrifugal pump are investigated. A method to systematically extract relevant frequencies that can be used as cavitation indicators is proposed. Various significant frequencies were found, distributed over the frequency range up to 30kHz. The cavitation on the impeller suction side was observed via a high-speed camera. A link between the optical appearance of the cavitation and the behavior of the indicator frequencies was apparent. Furthermore, a method for algorithmically evaluating the cavitation visible in high-speed imagery is presented and the synchronized optical and acoustical data was used to investigate the spectral change during transient cavitation events.