System and/or method for semantic parsing of air traffic control audio
The method S 200 can include: at an aircraft, receiving an audio utterance from air traffic control S 210 , converting the audio utterance to text, determining commands from the text using a question-and-answer model S 240 , and optionally controlling the aircraft based on the commands S 250 . The method functions to automatically interpret flight commands from the air traffic control (ATC) stream.
1. A system for an aircraft comprising:
an automatic speech recognition (ASR) system configured to determine a spelling hypothesis associated with a first waypoint entity based on an Air Traffic Control (ATC) audio signal;
a waypoint clarification system configured to:
in response to satisfaction of a trigger condition, automatically provide a spelling clarification request via an ATC radio; and
update the spelling hypothesis based on ASR analysis of the spelling clarification; and
a control system configured to control the aircraft based on a second waypoint entity associated with the updated spelling hypothesis.
2. The system of claim 1 , wherein the spelling hypothesis is based on a waypoint entity lexicon comprising the first waypoint entity and the second waypoint entity.
3. The system of claim 1 , wherein the trigger condition is based on at least one of:
a language score falling below a threshold value, wherein the language score is determined using a language model that is pretrained using waypoint pronunciations; or
a waypoint proximity relative to the aircraft which is determined using a prior map of waypoint entities.
4. The system of claim 3 , wherein the language model is refined based on the ATC audio signal and the ASR analysis of the spelling clarification.
5. The system of claim 1 , wherein the spelling hypothesis is determined with a language model, wherein the language model is updated based on the audio signal and the ASR analysis.
6. A method comprising:
based on an Air Traffic Control (ATC) audio signal, automatically determining a waypoint spelling hypothesis;
in response to satisfaction of a trigger condition, automatically requesting waypoint spelling clarification via an ATC radio communication system; and
determining a waypoint entity based on an analysis of an ATC response to the spelling clarification request; and
automatically determining an aircraft command based on the waypoint entity; and
facilitating control of an aircraft based on the aircraft command.
7. The method of claim 6 , wherein the waypoint spelling hypothesis is determined using a machine learning (ML) model, the method further comprising: updating the ML model based on the analysis of the ATC response to the spelling clarification request.
8. The method of claim 7 , wherein the analysis is determined using the ML model.
9. The method of claim 6 , further comprising: determining a language score associated with the waypoint spelling hypothesis, wherein the language score is determined using a language model that is pretrained using waypoint pronunciations, wherein satisfaction of the trigger condition is based on a language score falling below threshold value.
10. The method of claim 9 , wherein the language model is updated based on the analysis of the ATC response to the spelling clarification request.
11. The method of claim 10 , wherein the analysis comprises automatic speech recognition (ASR) using a pretrained neural network model.
12. The method of claim 11 , wherein the pretrained neural network model comprises the language model.
13. The method of claim 6 , wherein the waypoint spelling hypothesis or the satisfaction of the waypoint clarification condition is determined based on contextual information.
14. The method of claim 13 , wherein the contextual information comprises the ATC tower associated with the ATC audio signal.
15. The method of claim 13 , further comprising: receiving the contextual information prior to aircraft departure according to a flight plan.
16. The method of claim 6 , wherein the waypoint clarification is based on a language score falling below a threshold value, wherein the language score is determined using a language model that is pretrained using waypoint pronunciations.
17. The method of claim 6 , wherein the spelling hypothesis is based on a predetermined waypoint entity lexicon.
18. The method of claim 6 , wherein the waypoint spelling hypothesis is automatically determined by semantic parsing of ATC utterances.
19. The method of claim 6 , wherein the waypoint spelling hypothesis is associated with a semantic entity.
20. The method of claim 6 , wherein the trigger condition is based on waypoint proximity relative to the aircraft.