Pushing the Limits of Machine Translation with GPT-4 and Marian NMT in Multilingual Communication

Machine translation (MT) plays a crucial role in enabling linguistic diversity in online interactions. In this research, we provide a unique method for expanding the capabilities of MT inside a multilingual communication framework by combining GPT-4 with Marian Neural Machine Translation (NMT). We present a complete technique comprising translation quality evaluation, contextual coherence augmentation, and translation delay optimization. The results show that our method is successful in enhancing translation quality, preserving contextual coherence, and decreasing delay. Three algorithms are used in the suggested technique. 1) Contextual Coherence Enhancement, 2) Translation Latency Optimization, and 3) Translation Quality Assessment using BLEU and METEOR ratings. The first method uses BLEU and METEOR scores to quantitatively measure translation quality, guaranteeing linguistic precision. The second method improves conversational fluency by increasing contextual coherence. The third method minimizes the time it takes to translate a sentence, facilitating simultaneous translation and conversation. We evaluated our approach by comparing it to six established MT techniques. According to the outcomes, our strategy is superior to the status quo in many respects. In particular, our approach enhanced translation quality as measured by BLEU and METEOR ratings. It also had a far greater Contextual Coherence Enhancement Factor (CEF) and Coherence Improvement Percentage, indicating that it successfully preserved the rational progressions of conversations. Furthermore, significant improvements in translation latency were also highlighted by the Latency Optimization Ratio (LOR).