Compared to English language tasks, there is a greater demand for an efficient and successful automatic system for information retrieval and text summarization in Indonesian. Indonesian researchers try to find room to improve the solution for Indonesian text summarization task. Some of previous research for Indonesia text summarization which achieved good performance is using IndoBERT, the robust method in Indonesia text processing. Generally, Convolutional Neural Network (CNN) model is implemented in computer vision problem, however CNN in one dimension (1D CNN) is able to use for text classification because it can get local and hierarchical patterns in text data. This research tries to combine IndoBERT with one-dimensional convolutional neural network (1D CNN) using Liputan6 to conduct summarization using a dataset article Indonesia language. The IndoBERT step is used as a feature extraction and attention embedding article, one dimension CNN is then used to classify by determining the sentences included in the summary or not. ROUGE-1, ROUGE-2, and ROUGE-L are the metrics used in this study. The outcome demonstrates that the suggested approach performs better than the alternative model such as LEAD, ORACLE, and BERTEXT using the same metric evaluation.