Modulation recognition is a critical technique in signal processing, essential for analyzing and classifying the modulation schemes of received signals. This process is fundamentally important across various communication technologies, facilitating enhanced signal interpretation and network efficiency. This paper introduces a novel approach to modulation recognition that utilizes a combination of advanced neural network architectures. Specifically, the method integrates convolutional neural networks (CNNs), recurrent neural networks (RNNs), and attention mechanisms to form a comprehensive deep learning model that targets the intricate aspects of signal modulation. The proposed model leverages the spatial feature extraction capabilities of CNNs along with the sequential data processing strengths of RNNs, complemented by the focused insights provided by attention mechanisms. This synergy enhances the model’s ability to discern and classify complex modulation patterns more accurately than traditional methods. To validate the effectiveness of the proposed method, we conducted a series of rigorous experiments. These included detailed performance comparisons with existing modulation recognition techniques, assessments of the intra-class recognition abilities of our model, and analyses using partial confusion matrices to pinpoint areas of strength and potential improvement. The results from these experiments were highly encouraging. They clearly demonstrated that the proposed method not only achieves superior recognition rates but also excels in intra-class recognition tasks, where the ability to distinguish between similar types of modulation within the same class is critical. These outcomes underscore the potential of integrating CNNs, RNNs, and attention mechanisms in creating more robust and accurate modulation recognition systems, suggesting significant implications for future research and application in the field of communications.