In this study, an analytical model is proposed for natural frequency calculation and acoustic noise prediction for high speed switched reluctance machines. The developed natural frequency model results are compared with the mechanical finite element analysis results in terms of 6 different mode shapes that cause the majority of the acoustic noise in switched reluctance machines. The results show that the analytical results are consistent with the numerical method results with minimum 90% matching. Based on the natural frequency calculation model, a new acoustic noise prediction method is developed that only needs a radial force waveform as an input emerging on stator pole surfaces. The comparison of the developed and the numerical results clearly indicates that the acoustic noise level of the switched reluctance machine can be effectively found during the design process without using time-consuming numerical methods.