Can I read Decoder-Only LLMs for Speech-to-Text on EtoBox?

Decoder-Only LLMs for Speech-to-Text by revofor783 is a document available to read on EtoBox.

What is Decoder-Only LLMs for Speech-to-Text about?

This paper investigates the integration of decoder-only large language models (LLMs) into speech-to-text translation (S2TT), proposing a novel architecture that directly consumes encoded speech representations to generate text translations. The proposed model achieves state-of-the-art performance on benchmark datasets CoVoST 2 and FLEURS without relying on proprietary data, and the authors analyze various design choices to validate their approach. The research highlights the potential of LLMs in enhancing S

Author

revofor783

Language

EN