Transforming Speech Translation: An End-to-End Transformer-LAS-CTC System for Chinese Dialects to Minor Languages

Speech translation involves converting spoken language from one language to text in another. However, traditional speech translation systems use a cascaded structure that has several shortcomings and defects. As global trade and economic development continue to grow, there is an increasing demand for speech translation tasks from Chinese dialects to minor languages. To tackle these issues, this paper proposes an end-to-end transformer- Las-ctcspeech translation system for converting Chinese dialects to minor language text. Traditional systems have limitations, and with globalization, there is an increasing demand for speech translation. The proposed system combines the attention mechanism with a joint LAS-CTC decoder and outperforms traditional models in convergence speed and results. It has the potential to become a useful tool for bridging language barriers and facilitating communication.