An End-to-End Speech Summarization Using Large Language Model

TL;DR

This paper introduces an end-to-end speech summarization model leveraging large language models and multimodal fusion, effectively handling long speech inputs to generate concise summaries with competitive results.

Contribution

It proposes a novel multi-stage training framework using Q-Former and LLMs for direct speech-to-text summarization, addressing challenges of long input handling and cross-modal mapping.

Findings

Achieves competitive performance on How-2 dataset

Effectively handles long speech inputs

Utilizes curriculum learning for better transition from TSum to SSum

Abstract

Abstractive Speech Summarization (SSum) aims to generate human-like text summaries from spoken content. It encounters difficulties in handling long speech input and capturing the intricate cross-modal mapping between long speech inputs and short text summaries. Research on large language models (LLMs) and multimodal information fusion has provided new insights for addressing these challenges. In this paper, we propose an end-to-end SSum model that utilizes Q-Former as a connector for the audio-text modality and employs LLMs to generate text summaries directly from speech features. We adopt a multi-stage training approach that includes LLM based ASR and Text Summarization (TSum) tasks as auxiliary tasks. ASR tasks are used to align feature spaces and enhance the LLM's ability to handle longer speech. Then, we utilize a curriculum learning strategy to facilitate the model's transition from TSum to SSum. Finally, our model achieves competitive performance on the How-2 dataset.