LLaST: Improved End-to-end Speech Translation System Leveraged by Large Language Models

TL;DR

LLaST introduces a novel LLM-based speech translation framework that enhances end-to-end speech translation performance through architecture design and optimization, setting new benchmarks and scaling effectively.

Contribution

The paper presents a new LLM-based speech translation architecture with innovative training and optimization techniques, improving performance and scalability over existing models.

Findings

Superior performance on CoVoST-2 benchmark

Effective scaling capabilities with LLMs

Provides a strong baseline for future speech translation research

Abstract

We introduces LLaST, a framework for building high-performance Large Language model based Speech-to-text Translation systems. We address the limitations of end-to-end speech translation(E2E ST) models by exploring model architecture design and optimization techniques tailored for LLMs. Our approach includes LLM-based speech translation architecture design, ASR-augmented training, multilingual data augmentation, and dual-LoRA optimization. Our approach demonstrates superior performance on the CoVoST-2 benchmark and showcases exceptional scaling capabilities powered by LLMs. We believe this effective method will serve as a strong baseline for speech translation and provide insights for future improvements of the LLM-based speech translation framework. We release the data, code and models in https://github.com/openaudiolab/LLaST.