Command Extraction Main Algorithm

Abstract

Speech-to-text software transforms an analogue or digital speech signal into a sequence of words, e.g. “lufthansa one two bravo descend six thousand feet”. Text-to-concept, also known as command extraction, extracts the semantic meaning. It transforms the sequence of words from controller-pilot communication into ATC concepts being agreed in the SESAR funded solution 16-04. In the above example we obtain the callsign “DLH12B”, the type “DESCEND” the value “6000” and the unit “ft” as ATC concepts. DLR’s Insitute of Flight Guidance was the first to provide a software implementation for this task. This document describes the command extraction software for people who want to maintain the software Special focus is on a high level overview of the C++-Code of the main algorithms.