Spoken DialogSum: An Emotion-Rich Conversational Dataset for Spoken Dialogue Summarization

TL;DR

Spoken DialogSum is a new dataset linking speech, emotion, and summaries for spoken dialogue summarization, enabling emotion-aware models and improving summarization quality.

Contribution

It introduces the first emotion-rich spoken dialogue dataset with aligned speech, summaries, and paralinguistic labels, created through a novel multi-stage process involving LLM rewriting and speech synthesis.

Findings

Audio-LLM improves emotional-summary ROUGE-L by 28%.

Dataset enables emotion-aware dialogue summarization research.

Baseline models benefit from end-to-end speech modeling.

Abstract

Recent audio language models can follow long conversations. However, research on emotion-aware or spoken dialogue summarization is constrained by the lack of data that links speech, summaries, and paralinguistic cues. We introduce Spoken DialogSum, the first corpus aligning raw conversational audio with factual summaries, emotion-rich summaries, and utterance-level labels for speaker age, gender, and emotion. The dataset is built in two stages: first, an LLM rewrites DialogSum scripts with Switchboard-style fillers and back-channels, then tags each utterance with emotion, pitch, and speaking rate. Second, an expressive TTS engine synthesizes speech from the tagged scripts, aligned with paralinguistic labels. Spoken DialogSum comprises 13,460 emotion-diverse dialogues, each paired with both a factual and an emotion-focused summary. We release an online demo at https://fatfat-emosum.github.io/EmoDialog-Sum-Audio-Samples/, with plans to release the full dataset in the near future. Baselines show that an Audio-LLM raises emotional-summary ROUGE-L by 28% relative to a cascaded ASR-LLM system, confirming the value of end-to-end speech modeling.