Early Callsign Highlighting using Automatic Speech Recognition to Reduce Air Traffic Controller Workload

Abstract

The primary task of an air traffic controller (ATCo) is to issue instructions to pilots. However, the first contact is often initiated by the pilot. It is useful to have a controller assistance system, which could recognize and highlight the spoken callsign as early as possible, directly from the speech data. Therefore, we propose to use an automatic speech recognition (ASR) system to obtain the speech-to-text translation, using which we extract the spoken callsign. As a high callsign recognition performance is required, we use surveillance data, which significantly improves the performance. We obtain callsign recognition error rates of 6.2% and 8.3% for ATCo and pilot utterances, respectively, but can improve to 2.8% and 4.5%, when using information from surveillance data