Enhancing Automatic Speech Recognition in Air Traffic Communication Through Semantic Analysis and Error Correction

Effective communication is a critical component of successful human–machine interaction, particularly in safety–critical applications such as cockpit avionics speech. In these contexts, it is essential that both parties can understand what is being spoken and the context in which it was spoken. However, achieving high accuracy for automatic speech recognition remains a challenge in such applications. One of the contributing factors that affect the performance of ASR in air traffic communication is the domain vocabulary, which refers to the specific terminology and jargon used in the aviation industry. Additionally, the formation of phraseology, or the way phrases are constructed, and non-standard phraseology, which refers to non-standard or non-uniform phrasing used by different speakers, can also impact automatic speech recognition accuracy. To improve the accuracy in air traffic communication, this work focuses on incorporating semantic analysis to address error detection and correction. Semantic analysis involves understanding the meaning of language beyond just its literal interpretation, which can help to identify errors and inconsistencies in the spoken language. By combining error detection and correction with semantic analysis, the overall accuracy of the speech recognition system can be improved. The approach taken in this work leverages the semantic strength of the utterances used in air traffic communication to enhance the accuracy of speech recognition. By analyzing the meaning behind the words and phrases used, errors can be more easily identified and corrected. This helps to ensure that the communication between the human operator and the machine is as accurate and effective as possible.