The rising prevalence of mental health challenges underscores the need for scalable, accessible, and personalized interventions. Emotion Recognition Technology (ERT) offers a promising solution by integrating artificial intelligence and computational psychology to detect, interpret, and respond to individual emotional states in real-time. This research introduces an Emotion Recognition Technology (ERT) framework designed to deliver personalized digital mental health support interventions. The framework was evaluated with 125 participants, and through iterative testing, each user’s most reliable modality, facial expressions or vocal tone, was identified to capture emotional states with higher accuracy. Facial data were preprocessed through cropping, while speech data underwent bandpass filtering. FaceNet was applied to extract features from facial image data, and Mel-Frequency Cepstral Coefficients (MFCCs) were used for speech data. A weighted fusion algorithm combined predictions across modalities to refine overall emotion detection, ensuring personalized forecasts tailored to individual user profiles. To translate emotion recognition into meaningful support, the system employs an Efficient Dandelion Optimizer–driven Attention-refined Variational Autoencoder (ED-Att-VAE). This model recommends Cognitive Behavioral Therapy (CBT) activities validated by mental health professionals. Activities such as guided relaxation, journaling prompts, and cognitive reframing tasks are delivered through a mobile health (mHealth) platform and dynamically adapted to user emotions. Results show that the proposed model, implemented in Python, achieved a high performance with an accuracy of 97.2%. The effectiveness of personalized interventions was assessed using the DASS-21 questionnaire, where the experimental group demonstrated significant reductions in depression and anxiety scores compared to the control group. These findings highlight the potential of combining deep learning–based multimodal emotion prediction with CBT personalization to enhance mental well-being. This research contributes to the advancement of emotion-aware, personalized digital mental health interventions, bridging the gap between technological innovation and clinically validated care.