Adapting Automatic Speech Recognition for Accented Air Traffic Control Communications

2025-02-27

AI Summary

To address the low automatic speech recognition (ASR) accuracy for Southeast Asian-accented English in noisy air traffic control (ATC) environments, this work introduces the first accent-specific ATC speech dataset tailored to the region and proposes an “accent-focused, region-adapted” ASR training paradigm. Methodologically, we integrate domain-adaptive fine-tuning of the Whisper architecture, noise-robustness enhancement, accent-aware feature alignment, and lightweight inference optimization—balancing recognition performance with deployment constraints in resource-limited military scenarios. Evaluated on our proprietary test set, the proposed system achieves a word error rate (WER) of 9.82%, substantially outperforming generic ASR models. This demonstrates the efficacy of region-specific modeling in improving robustness to linguistic variation under high-noise conditions. The study establishes a reproducible technical framework for multilingual, accented, and noise-robust professional speech recognition.

Technology Category

• Natural Language Processing: Speech
• Application Domains: Transportation
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Large-scale security measurements

Abstract

Effective communication in Air Traffic Control (ATC) is critical to maintaining aviation safety, yet the challenges posed by accented English remain largely unaddressed in Automatic Speech Recognition (ASR) systems. Existing models struggle with transcription accuracy for Southeast Asian-accented (SEA-accented) speech, particularly in noisy ATC environments. This study presents the development of ASR models fine-tuned specifically for Southeast Asian accents using a newly created dataset. Our research achieves significant improvements, achieving a Word Error Rate (WER) of 0.0982 or 9.82% on SEA-accented ATC speech. Additionally, the paper highlights the importance of region-specific datasets and accent-focused training, offering a pathway for deploying ASR systems in resource-constrained military operations. The findings emphasize the need for noise-robust training techniques and region-specific datasets to improve transcription accuracy for non-Western accents in ATC communications.

Problem

• Improving ASR for accented ATC communications
• Enhancing transcription accuracy for SEA-accented speech
• Developing noise-robust ASR models for military operations

Innovation

• ASR models fine-tuned
• Southeast Asian accents
• noise-robust training techniques