ACI-BENCH variant

This repository contains the original ACI-Bench dataset, along with an ASR-degraded variant created to simulate real-world automatic speech recognition (ASR) errors in clinical conversation transcripts. This data was used to evaluate the effectiveness of open-source language models in automatically generating patient assessment summaries from uncorrected ASR transcriptions of doctor-patient conversations. The research includes the consideration of the impact of transcription errors on the quality of the generated summaries.