In the current world, with remarkable advances in information technology, text data is rapidly increasing and to access information in a variety of areas, individuals are faced with a massive amount of documents and texts from different sources. Thus, many researches have been commenced to develop and provide text summarization methods for faster and easier access to documents. Among the methods that the text summarization is performed by human intervention, machine learning methods tend to focus on improving the quality of features and minimizing human interference in the process of selecting sentences. Since the deep learning methods have improved the performance of the tasks related to issues such as image processing, natural language processing, and so on, hence these methods were used to enhance the performance of the tasks related to text processing too. In this study, we present a new model for producing extracted text summaries using deep learning and clustering techniques. Our proposed model is based on the Gaussian- Bernoulli Restricted Boltzmann Machine, Deep Belief Networks, and k-means clustering algorithm. In this method, we first select several proper features of the text sentences and then we construct a feature matrix. The feature matrix is applied it to a Restricted Boltzmann machine and a Deep Belief Network that have been improved to be applicable for the real inputs. Also, few improvements have been made to the k-means clustering algorithm, as a complement to the methods of deep learning for choosing important sentences of the text. These improvements caused producing high-quality summaries that include key sentences representing the concept of a text document. To evaluate the summaries produced by the proposed method in comparison with a human-made summary, the DUC2002 text data is used. The proposed method based on the Deep Belief Networks and the clustering method achieved average 51.33% for the ROUGE-1 evaluation criteria and 24.45% for the ROUGE-2 evaluation criteria. The experimental result shows that the summaries generated using the proposed method has good quality and has better performance in comparison with other similar text summarization methods. Key Words: Text Summarization, Deep learning, Machine learning, Clustering, Restricted Boltzmann Machine, Deep Belief Network.