Train timetable plays a crucial role in railway transportation operations, outlining the arrival and departure times for each train. The precision of the timetable is vital in ensuring the efficiency, stability, and safety of train operations. In this paper, we introduce a precision-oriented approach to generating train timetables, leveraging a space-time network framework. An integer programming model is formulated to capture the complexities of timetable generation, and an efficient alternating direction method of multipliers (ADMM) decomposition algorithm is designed to solve the model. Utilizing the lower bound solution obtained through the ADMM-based decomposition algorithm, we further propose a topological order-based algorithm to generate an upper bound solution, resulting in a feasible timetable with second-level granularity. Finally, we validate the effectiveness and efficiency of our proposed model and algorithm through a real-world experiment involving the Beijing-Shanghai high-speed railway.