Generating Synthetic Doctor-Patient Conversations for Long-form Audio Summarization

TL;DR

This paper introduces a synthetic data pipeline for long-form audio reasoning, focusing on doctor-patient conversations and SOAP note generation, to improve training and evaluation of audio understanding systems.

Contribution

It presents a novel multi-stage pipeline for generating synthetic long-context audio conversations and reference notes using open-weight models, addressing data scarcity and evaluation challenges.

Findings

Cascaded systems outperform end-to-end models on synthetic data.

8,800 synthetic conversations and 1.3k hours of audio are released.

The pipeline enables controlled evaluation of long-context audio reasoning.

Abstract

Long-context audio reasoning is underserved in both training data and evaluation. Existing benchmarks target short-context tasks, and the open-ended generation tasks most relevant to long-context reasoning pose well-known challenges for automatic evaluation. We propose a synthetic data generation pipeline designed to serve both as a training resource and as a controlled evaluation environment, and instantiate it for first-visit doctor-patient conversations with SOAP note generation as the task. The pipeline has three stages, persona-driven dialogue generation, multi-speaker audio synthesis with overlap/pause modeling, room acoustics, and sound events, and LLM-based reference SOAP note production, built entirely on open-weight models. We release 8,800 synthetic conversations with 1.3k hours of corresponding audio and reference notes. Evaluating current open-weight systems, we find that cascaded approaches still substantially outperform end-to-end models.