This paper proposes a novel deep learning model named RTEU-Transformers (Residual Temporal Enhancement Unit) for abstractive text summarization, specifically applied to the Samsum dataset, which consists of human-to-human dialogue transcripts. Unlike conventional Transformer-based models that often struggle with conversational context retention and temporal coherence, RTEU-Transformers introduce a Residual Temporal Enhancement Unit that captures the progression of dialogue turns and reinforces important temporal dependencies. By integrating residual connections and hierarchical temporal encoding, our model enhances semantic preservation and summary fluency. Performance evaluation was conducted using ROUGE-1, ROUGE-2, and ROUGE-L metrics. RTEU-Transformers achieved ROUGE-1: 51.7, ROUGE-2: 26.5, and ROUGE-L: 47.8, surpassing existing baselines such as BART, T5, and PEGASUS. Compared to BART (ROUGE-1: 48.5), our model shows significant improvements in capturing contextual subtleties of dialogues. The novelty of this approach lies in enhancing dialogue understanding with minimal computational overhead. The proposed model not only boosts summarization accuracy but also demonstrates better generalization across unseen dialogues. This paper contributes a step forward in improving conversational AI systems and dialogue-based summarization, with implications for customer service automation, messaging apps, and assistive tools.