Software defect prediction plays a crucial role in the improvement of decreasing maintenance costs and the quality of the software. Traditional machine learning (ML) models usually cannot perform better in every aspect at a time for Speed, generalization, and accuracy. To enhance the performance of Extreme Learning Machines (ELM), we developed an ensemble-based model. It uses an approach that combines the Dolphin Swarm Algorithm and Bat Algorithm to optimize the parameters of the ELM. Prediction accuracy and convergence are improved by this proposed model. NASA datasets are used to evaluate the performance of the proposed hybrid model. The proposed hybrid model outperforms the conventional models.