DeRAGEC: Denoising Named Entity Candidates with Synthetic Rationale for ASR Error Correction

2025-06-09

AI Summary

This work addresses the low accuracy of named entity (NE) correction and severe interference from noisy candidates in automatic speech recognition (ASR) systems. We propose a training-free, synthetic-reasoning-driven denoising mechanism within a retrieval-augmented generation for error correction (RAGEC) framework. Our method jointly models phonetic similarity and knowledge-enhanced entity definitions to automatically generate interpretable synthetic rationales, enabling in-context zero-shot pre-filtering and correction of NE candidates. The core contribution is the first integration of phonetic similarity with structured entity definitions for synthetic rationale generation—eliminating reliance on fine-tuning while significantly reducing noise interference. Evaluated on CommonVoice and STOP datasets, our approach reduces word error rate (WER) by 28% over baseline ASR systems and substantially improves NE recall.

Technology Category

• Natural Language Processing: Generation
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Machine Learning: Neuro-Symbolic Learning

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We present DeRAGEC, a method for improving Named Entity (NE) correction in Automatic Speech Recognition (ASR) systems. By extending the Retrieval-Augmented Generative Error Correction (RAGEC) framework, DeRAGEC employs synthetic denoising rationales to filter out noisy NE candidates before correction. By leveraging phonetic similarity and augmented definitions, it refines noisy retrieved NEs using in-context learning, requiring no additional training. Experimental results on CommonVoice and STOP datasets show significant improvements in Word Error Rate (WER) and NE hit ratio, outperforming baseline ASR and RAGEC methods. Specifically, we achieved a 28% relative reduction in WER compared to ASR without postprocessing. Our source code is publicly available at: https://github.com/solee0022/deragec

Problem

• Improves Named Entity correction in ASR systems
• Filters noisy NE candidates using synthetic denoising rationales
• Enhances WER and NE hit ratio without additional training

Innovation

• Synthetic denoising rationales filter NE candidates
• Phonetic similarity refines noisy retrieved NEs
• In-context learning requires no additional training