With the rapid development of the Internet, there is a large amount of conflicting data on the network. It is becoming increasingly important to identify the true information. Truth discovery becomes a hot research topic, it identifies object truth while estimating source reliability. Although many methods have been proposed, they are all applicable to specific scenarios, and none of them can be applied to all scenarios or always better than others. That is, there is no one-fit-all solution. In addition, objects have different difficulties. Accurately assessing the difficulty of each object is conducive to improving the accuracy of the truth discovery. In order to solve the above problem, we propose a distributed ensemble truth discovery model (DETD), which combines different results output by base truth discovery algorithms, while considering the difficulty of each object and computing the reliability of base algorithms. Experimental results show that the proposed DETD model can effectively estimate the truths of objects on two real-world datasets.