Synthetic Audio Generation Framework for Air Traffic Control Speech Recognition

2026-06-19

AI Summary

This study addresses the performance limitations of air traffic control (ATC) automatic speech recognition (ASR) systems caused by high levels of background noise, non-native accents, and scarcity of real-world training data. To overcome these challenges, the authors propose a high-fidelity synthetic speech generation framework tailored to ATC scenarios, which innovatively incorporates a controllable L1-to-L2 accent transformation method. By integrating text-to-speech synthesis, voice conversion, and accent transfer techniques, the framework effectively simulates realistic acoustic conditions and accent variations. Fine-tuning Whisper models with data generated by this framework—using either purely synthetic or hybrid datasets—yields substantial reductions in word error rate on the ATCO2 corpus, significantly outperforming both off-the-shelf models and baselines trained solely on real data. These results demonstrate the efficacy and potential of synthetic data for low-resource ATC ASR.

Abstract

Automatic Speech Recognition (ASR) systems, despite achieving remarkable accuracy in general-purpose domains with native speech (L1), struggle in domains like Air Traffic Control (ATC) due to strong channel noise, a presence of non-native (L2) English accents, and data scarcity. We propose a synthetic data generation pipeline with acoustical properties simulations specifically designed to address this lack of real data to improve recognition accuracy in the ATC domain. Our approach leverages a combination of neural generation techniques, including Text-to-Speech, Voice Conversion, L2-to-L1 accent conversion, and a novel controllable L1-to-L2 accent conversion framework built to simulate accented speech. Our experiments with the Whisper model on the ATCO2 corpus demonstrate that fine-tuning with either synthetic data alone, or a mix of real and synthetic data, significantly improves the word error rate over out-of-the-box and real data only baselines respectively.

Problem

• Air Traffic Control
• Automatic Speech Recognition
• Non-native Accents
• Data Scarcity
• Channel Noise

Innovation

• synthetic audio generation
• accent conversion
• air traffic control speech recognition
• data scarcity mitigation
• neural voice synthesis