Emotion recognition through brain computer interfaces has become increasingly important for applications in healthcare, human–computer interaction, and affective computing. Emotion recognition through brain–computer interfaces has become increasingly important for applications in healthcare, human–computer interaction, and affective computing. However, the high dimensionality, subject variability, and non linear characteristics of EEG signals present significant challenges for robust emotion classification. An Explainable Artificial Intelligence based Emotion(X-EMO) Recognition Framework is a deep learning model designed to recognize and interpret emotional states from EEG signals, which provides transparency and interpretability in its decision making process.This paper proposes an X EMO Inspired Multimodal Deep Learning Framework with Channel Specific Attention (CSA) for EEG based emotion recognition. The framework integrates convolutional and recurrent neural architectures to extract both spatial and temporal dependencies from multi channel EEG signals. A channel specific attention mechanism is employed to adaptively assign weights to informative electrode channels, enabling the model to focus on the most emotionally relevant brain regions. Additionally, multimodal fusion of EEG features with complementary physiological or contextual information further enhances recognition accuracy and generalization. Experimental evaluation demonstrates that the proposed framework achieves an overall accuracy of 98%, significantly outperforming conventional CNN based 85.2%, RNN based 86.7%, and hybrid CNN and RNN 89.3% models. The results highlight the effectiveness of channel aware multimodal learning for developing reliable, explainable, and real time EEG emotion recognition systems. The high dimensionality of multi channel EEG signals, subject variability, non linear dynamics, and noisy measurements make robust emotion classification extremely difficult using conventional machine learning methods. Moreover, most existing deep learning approaches treat all EEG channels equally, failing to focus on brain regions that are most informative for emotion recognition, which limits both accuracy and interpretability.