In this study, we introduce the 'CNC Assistant' for machining parameters selection for high speed requirement in accuracy and surface roughness constraints. It involves modeling for experimental data of accuracy, surface roughness and machining time on a five-axis CNC machine tool (HM3025L) published by CHMER, coupled with a controller (M6HN) which is launched by GENTEC, based on processing parameters (feed rate, acceleration after interpolation time constant, acceleration and S-curve time constant). In order to predict and optimize the processing parameters combination, we use the data-driven approach to establish the back-propagation neural network (BPNN), and apply the particle swarm optimization (PSO) algorithm to search the processing parameters based on the constraints of accuracy and surface roughness. That is, users can set the specified conditions of accuracy and surface roughness, then the CNC assistant has the ability to obtain the corresponding machining parameters, not only leading to the shortest machining time but also meeting the design conditions. As above, the CNC assistant provide the machinery industry become more intelligent and convenient to improve the efficiency of CNC machine tools.