There is an urgent demand for free form products in industry at the present time because of their superior appearance and the wide variety of functions they perform. Five-axis high-speed CNC machining technology has developed to satisfy this demand, but further improvement in surface quality metric inspection technology is the big challenge it now faces. In this study, the effects of jerk on the performance of five-axis synchronous high-speed CNC ball nose end mills on a freeform turbine mold were investigated. The relationships of characteristics of the images of 14 jerk-cluster finished workpieces with different jerk setting values were established, allowing surface texture features to be analyzed and surface roughness predicted. In addition, machine learning methods were integrated with the surface feature analysis to construct a virtual machining module that acts as a performance prediction system, merging the virtual machine tool functions, surface texture processor and AI roughness prediction processor. Using the geometric information of the workpiece, cutting parameters and machine tool parameters as inputs, product performance metrics combining surface roughness and machining time can be predicted as outputs of the system. The integrated system provides users with a way to evaluate manufacturing performance before performing actual operations and to reduce test time for cutting parameter development. The model is suitable for complex surface finishes as well as for the production of small batches with high parametric variance. In addition, the partial set of image processing and roughness prediction modules can be used alone as an effective intelligent surface quality inspection system.