Prototype of Smartglasses English Translator in the LSTM-Based Tourism Sector and Speech Recognition and Speech Synthesis Technology

Language barriers remain a significant challenge in the tourism sector, particularly in Indonesia, where communication between international tourists and local service providers often faces difficulties due to limited English proficiency. Existing translation tools, such as mobile apps and standalone devices, suffer from limitations like reliance on internet connections, one-way communication, and the need for additional devices. This research aims to develop a smart glasses prototype that integrates speech recognition, Long Short-Term Memory (LSTM)-based translation, and speech synthesis technologies to offer real-time, bidirectional communication. The prototype development involved several stages: literature review, design, tool collection, prototyping, and testing using the Tatoeba dataset, which includes 13,570 English-Indonesian sentence pairs. The smart glasses prototype was tested for translation accuracy, speech recognition, and synthesis performance, yielding BLEU scores of 36.5% for Indonesian-English and 39.8% for English-Indonesian, with accuracy rates of 80.8% and 87.8%, respectively. The prototype demonstrated effective real-time translation with speech-to-text and text-to-speech functionalities. These results show that the smart glasses prototype can significantly improve the quality of communication between tourists and local communities, enhancing the competitiveness of Indonesia's tourism industry. The study implies that wearable translation devices could become essential tools in tourism, reducing language barriers and fostering better cross-cultural interactions.