Automatic speech recognition for air traffic control communications

Abstract

Air Traffic Control (ATC) plays a crucial role in ensuring the safe and efficient passage of aircraft through the airspace. Unfortunately, miscommunications between pilots and air traffic controllers are a leading cause of aviation accidents. As most of the pilot-controller communications are conducted through speech via radio channels, this paper investigates the potential of automatic speech recognition (ASR) technology to enhance communications between pilots and air traffic controllers by reducing errors and improving efficiency. This paper focused on developing an effective ASR model tailored to the ATC domain. The model is also used to extract operational information such as call signs, radio frequencies, heading information, and flight levels. Utilising current machine learning and natural language processing technologies, a robust ASR model wastrained and developed. This paper outlines the entire process of creating the ASR model and compares the effect of using different machine learning techniques on achieving the optimal model. Lastly, the paper discusses the challenges and limitations faced during the development of the ASR model and potential future work that could be done to improve the model. Overall, this paper provides a solution to improve air traffic control using automatic speech recognition technology and provides a comprehensive overview of the steps involved in creating an effective ASR model for the ATC domain.