There have been two recent studies on the microscopic heat engines, one of which is a Brownian Carnot engine [1] and the other is an active Stirling engine [2]. The former is driven by Gaussian white noises and the latter is driven by bacteria. The active Stirling engine shows an efficiency higher than that limited by the macroscopic Stirling engine known in the traditional thermodynamics. It raises the question which properties of the noises are crucial for the high efficiency? Is it the temporal correlation or the magnitude statistics, such as the Gaussian or non-Gaussian distribution? To clarify these questions, we analytically and numerically study the effects of various external noises on several thermodynamic quantities of a Stirling heat engine, such as the work, the heat, and the efficiency. These stochastic driving sources include oscillatory noises and different Gaussian colored noises. These results provide a theoretical basis for the search of high efficient microscopic heat engines in experimental studies.