Saved in:
Bibliographic Details
Main Authors: Kumar, Anurag, Paturi, Rohit, Afshan, Amber, Srinivasan, Sundararajan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.08421
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916566763831296
author Kumar, Anurag
Paturi, Rohit
Afshan, Amber
Srinivasan, Sundararajan
author_facet Kumar, Anurag
Paturi, Rohit
Afshan, Amber
Srinivasan, Sundararajan
contents Speaker Diarization (SD) is a crucial component of modern end-to-end ASR pipelines. Traditional SD systems, which are typically audio-based and operate independently of ASR, often introduce speaker errors, particularly during speaker transitions and overlapping speech. Recently, language models including fine-tuned large language models (LLMs) have shown to be effective as a second-pass speaker error corrector by leveraging lexical context in the transcribed output. In this work, we introduce a novel acoustic conditioning approach to provide more fine-grained information from the acoustic diarizer to the LLM. We also show that a simpler constrained decoding strategy reduces LLM hallucinations, while avoiding complicated post-processing. Our approach significantly reduces the speaker error rates by 24-43% across Fisher, Callhome, and RT03-CTS datasets, compared to the first-pass Acoustic SD.
format Preprint
id arxiv_https___arxiv_org_abs_2501_08421
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SEAL: Speaker Error Correction using Acoustic-conditioned Large Language Models
Kumar, Anurag
Paturi, Rohit
Afshan, Amber
Srinivasan, Sundararajan
Audio and Speech Processing
Artificial Intelligence
Computation and Language
Machine Learning
Sound
Speaker Diarization (SD) is a crucial component of modern end-to-end ASR pipelines. Traditional SD systems, which are typically audio-based and operate independently of ASR, often introduce speaker errors, particularly during speaker transitions and overlapping speech. Recently, language models including fine-tuned large language models (LLMs) have shown to be effective as a second-pass speaker error corrector by leveraging lexical context in the transcribed output. In this work, we introduce a novel acoustic conditioning approach to provide more fine-grained information from the acoustic diarizer to the LLM. We also show that a simpler constrained decoding strategy reduces LLM hallucinations, while avoiding complicated post-processing. Our approach significantly reduces the speaker error rates by 24-43% across Fisher, Callhome, and RT03-CTS datasets, compared to the first-pass Acoustic SD.
title SEAL: Speaker Error Correction using Acoustic-conditioned Large Language Models
topic Audio and Speech Processing
Artificial Intelligence
Computation and Language
Machine Learning
Sound
url https://arxiv.org/abs/2501.08421