A closed-loop brain-machine interface requires the integration of the neural stimulation and recording circuits with the microelectrode array. A challenge for simultaneous neural recording and stimulation is the large interference faced by the neural amplifier induced by stimulation artifacts. In this work, front-end circuit design technique and digital post-processing are combined to provide a solution for fast recovery of the neural amplifier from artifact, based on a signal folding scheme. A chip integrating the neural recording and stimulation circuits is implemented in GF-0.18μm CMOS process for proof of concept. In - vitro experiment is conducted using this chip, showing the viability of the proposed design. At stimulation current levels from 50 - 100 μA, which is sufficient for deep brain stimulation, a faster recovery time of around 1 ms is achieved with the proposed neural amplifier compared to 4 ms from conventional one.