About this document
S2ST with LLMs: Feasibility Study by Niley Gonzalez is a document available to read on EtoBox.
The document discusses the feasibility and alignment strategies of leveraging Large Language Models (LLMs) for Speech to Speech Translation (S2ST). It outlines the cascade and direct S2ST approaches, highlighting the advantages of LLMs in handling multiple speech tasks and facilitating translation for languages without written forms. The study emphasizes the use of prompting techniques for effective learning in speech tasks and the transfer of linguistic knowledge from text to speech.
- Author
- Niley Gonzalez
- Language
- EN