Bearing is a rotating machine element that keeps the machine performance in good condition. Bearing enables the shaft to rotate without excessive friction. Bearing faults may reduce machine performance, stop operation machine, decrease production, and increase maintenance costs. The purpose of this research was to detect bearing fault inner race and outer race based on vibration analysis using frequency domain and envelope spectrum. Vibration analysis is the most popular and widely used method in CBM (Condition Based Maintenance) especially for analyzing bearing condition with features such as frequency domain and envelope analysis. Measurements were made on a simple model of rotor shaft system. The bearing on the system was divided into two conditions, bearings with multi-type defects and normal bearing conditions with 4 shaft speed variation (1000 RPM, 1200 RPM, 1400 RPM, 1600 RPM). Bearing used Self-aligning ball bearing type, SKF, 1207, 1207 EKTN9/ C3 Series. Multi-type bearing faults were inner race and outer race. In frequency domain, normal bearing did not show the frequency of bearing faults. The multi-type faults condition on bearing occur 1X to 2X harmonic, but frequency which had low amplitude value with the frequency of other components that were around the bearing. The envelope analysis method caused the impacts with very low energy. It also brought up the frequency amplitude of bearing faults that closed or immersed in frequency domain analysis. Multi-type faults condition on bearing of each shaft speed variation showed the frequency of inner race faults (BPFI) and outer race faults (BPFO) with different amplitude values. The higher speed of shaft caused the higher amplitude of faulty bearing frequency.