Real-Time Accent Translation

This literature review provides a comprehensive analysis of existing research in the field of accent conversion and its application to real-time speech translation. The focus is on non-autoregressive models, which offer low-latency accent conversion, a critical requirement for real-time applications. The review covers key advancements in phonetic-based accent conversion, zero-shot learning methods, and neural network architectures that enable efficient accent translation. Finally, the review highlights the limitations of traditional methods and emphasizes the role of Nechaev and Kosyakov’s non-autoregressive real-time accent conversion model as a foundation for future developments in this area.