There are number of different methods and procedures in vibration analysis, where the natural frequencies of the specimen or the system are one of the key parameters. It is known that these frequencies can change under load, for example in response to pre-stressing, but the effect of residual stresses is less known. By developing a suitable method, natural frequencies can be used to predetermine residual stress, therefore this method can be used for example predicting whether it will cause deformation during machining of a part, whether it requires increased attention or how to set the parameters well for vibratory stress relief. The results can be significant cost and time savings, as well as the improvements of the quality. Natural frequency is the frequency of free vibration of an undamped linear vibration system, or in other words at which a system left alone will vibrate after excited by an external force [1]. Metal castings or welded structures may have several natural frequencies which appear as frequency bands or ranges on the measurement images. Based on these, to determine the natural frequency of a component or system, we need to excite a frequency as close as possible to the natural frequency for the resonance to occur. When the resonance is reached, the amplitude of the system is at its maximum, and the natural frequencies of the workpiece can be measured. Traditionally, sensors, usually accelerometers are used to measure the natural frequency. The continuous development of information technology has made it possible to replace these sensors with an acoustic diagnostic system. During this research, we have developed an acoustic diagnostic system and procedure, which can generate the acoustic measurement images. We have evaluated the measurement images in many ways, and many different types of components and materials (mostly iron alloys) were analyzed. In addition, the changes of natural frequencies show a similar pattern in the case of parts before treating with vibratory stress relief as for load tests.