In this study, We use microstep control to reduce vibration of step motor. Microstep control of step motor is usually thought of as an extension of conventional step motor control technology. The essence of microstepping is that we divide the full step of a step motor into a number of substep called microstep and cause the stepmotor to move through a substep per input pulse. In ideal case, by controlling the individual phase currents of a two-phase step motor sinusoidally we can get uniform torque and step angle. But due to the nonlinear characteristics of the step motor, we need to compensate current waveform to improve the overall smoothness of the conventional micro stepping system. We implement digital Pulse Width Modulation(PWM) driver to drive step motor and microphone was used fordetecting vibration. Driver enables speed change automatically by increasing or decreasing micro stepping ratio which we call Automatic Switching on th Fly. To compensate the torque harmonics, neural network is applied to the system and we found compensated optimal input current waveform. Finally we can get smooth motion of step motor in a wide range of motor speed.