Adaptive speech recognition methods and systems
Methods and systems are provided for assisting operation of a vehicle using speech recognition. One method involves analyzing a transcription of an audio communication with respect to the vehicle to characterize a nonstandard pattern within the transcription of the audio communication, obtaining a ground truth for the transcription of the audio communication, determining one or more performance metrics associated with the nonstandard pattern within the transcription based on a relationship between the transcription of the audio communication and the ground truth for the transcription, updating a speech recognition vocabulary for the vehicle to include the nonstandard pattern based at least in part on the one or more performance metrics and determining an updated speech recognition model for the vehicle using the updated speech recognition vocabulary and the audio communication.
1. A method of assisting operation of a vehicle, the method comprising:
analyzing a transcription of an audio communication with respect to the vehicle to characterize a nonstandard pattern within the transcription of the audio communication;
obtaining a ground truth for the transcription of the audio communication;
determining one or more performance metrics associated with the nonstandard pattern within the transcription based on a relationship between the transcription of the audio communication and the ground truth for the transcription;
updating a speech recognition vocabulary for the vehicle to include the nonstandard pattern based at least in part on the one or more performance metrics, resulting in an updated speech recognition vocabulary; and
determining an updated speech recognition model for the vehicle using the updated speech recognition vocabulary and the audio communication.
2. The method of claim 1 , further comprising pushing the updated speech recognition model to the vehicle over a communications network.
3. The method of claim 2 , further comprising analyzing, at the vehicle, a second transcription of a subsequent audio communication with respect to the vehicle to detect the nonstandard pattern within the second transcription of the subsequent audio communication using at least one of the updated speech recognition vocabulary and the updated speech recognition model.
4. The method of claim 1 , further comprising analyzing a second transcription of a subsequent audio communication with respect to the vehicle to detect the nonstandard pattern within the second transcription of the subsequent audio communication using at least one of the updated speech recognition vocabulary and the updated speech recognition model.
5. The method of claim 1 , wherein analyzing the transcription of the audio communication with respect to the vehicle to characterize the nonstandard pattern within the transcription of the audio communication comprises:
determining at least one of a phraseology pattern subject category and a phraseology pattern structure type associated with the transcription of the audio communication based at least in part on content of the transcription of the audio communication;
determining whether the content of the transcription of the audio communication corresponds to one of a plurality of standard phraseology patterns based at least in part on the at least one of the phraseology pattern subject category and the phraseology pattern structure type; and
determining the transcription of the audio communication comprises the nonstandard pattern when the transcription of the audio communication does not correspond to any of the plurality of standard phraseology patterns.
6. The method of claim 5 , further comprising assigning a pattern identifier associated with an existing nonstandard pattern to the transcription of the audio communication when the content of the transcription of the audio communication corresponds to the existing nonstandard pattern based at least in part on the at least one of the phraseology pattern subject category and the phraseology pattern structure type.
7. The method of claim 1 , wherein:
analyzing the transcription of the audio communication with respect to the vehicle to characterize the nonstandard pattern comprises identifying a phraseology pattern portion associated with the nonstandard pattern within the transcription of the audio communication; and
determining the one or more performance metrics associated with the nonstandard pattern within the transcription comprises determining a phraseology pattern performance metric based on a relationship between the phraseology pattern portion of the transcription of the audio communication and a second phraseology pattern portion of the ground truth.
8. The method of claim 1 , wherein:
determining the one or more performance metrics comprises determining a pattern-based performance metric associated with the nonstandard pattern within the transcription based on a relationship between a phraseology pattern portion of the transcription of the audio communication and the phraseology pattern portion of the ground truth for the transcription; and
updating the speech recognition vocabulary comprises updating the speech recognition vocabulary based on the pattern-based performance metric.
9. A non-transitory computer-readable medium having computer-executable instructions stored thereon that, when executed by a processing system, cause the processing system to:
analyze a transcription of an audio communication with respect to a vehicle to characterize a nonstandard pattern within the transcription of the audio communication;
obtain a ground truth for the transcription of the audio communication;
determine one or more performance metrics associated with the nonstandard pattern within the transcription based on a relationship between the transcription of the audio communication and the ground truth for the transcription;
update a speech recognition vocabulary to include the nonstandard pattern based at least in part on the one or more performance metrics, resulting in an updated speech recognition vocabulary; and
determine an updated speech recognition model for the vehicle using the updated speech recognition vocabulary and the audio communication.
10. The non-transitory computer-readable medium of claim 9 , wherein the computer-executable instructions cause the processing system to push the updated speech recognition model to the vehicle over a communications network.
11. The non-transitory computer-readable medium of claim 9 , wherein the computer-executable instructions cause the processing system to analyze a second transcription of a subsequent audio communication with respect to the vehicle to detect the nonstandard pattern within the second transcription of the subsequent audio communication using at least one of the updated speech recognition vocabulary and the updated speech recognition model.
12. The non-transitory computer-readable medium of claim 9 , wherein the computer-executable instructions cause the processing system to analyze the transcription of the audio communication with respect to the vehicle to characterize the nonstandard pattern within the transcription of the audio communication by:
determining at least one of a phraseology pattern subject category and a phraseology pattern structure type associated with the transcription of the audio communication based at least in part on content of the transcription of the audio communication;
determining whether the content of the transcription of the audio communication corresponds to one of a plurality of standard phraseology patterns based at least in part on the at least one of the phraseology pattern subject category and the phraseology pattern structure type; and
determining the transcription of the audio communication comprises the nonstandard pattern when the transcription of the audio communication does not correspond to any of the plurality of standard phraseology patterns.
13. The non-transitory computer-readable medium of claim 12 , wherein the computer-executable instructions cause the processing system to assign a pattern identifier associated with an existing nonstandard pattern to the transcription of the audio communication when the content of the transcription of the audio communication corresponds to the existing nonstandard pattern based at least in part on the at least one of the phraseology pattern subject category and the phraseology pattern structure type.
14. The non-transitory computer-readable medium of claim 9 , wherein:
analyzing the transcription of the audio communication with respect to the vehicle to characterize the nonstandard pattern comprises identifying a phraseology pattern portion associated with the nonstandard pattern within the transcription of the audio communication; and
determining the one or more performance metrics associated with the nonstandard pattern within the transcription comprises determining a phraseology pattern performance metric based on a relationship between the phraseology pattern portion of the transcription of the audio communication and a second phraseology pattern portion of the ground truth.
15. The non-transitory computer-readable medium of claim 9 , wherein the computer-executable instructions cause the processing system to:
determine a pattern-based performance metric associated with the nonstandard pattern within the transcription based on a relationship between a phraseology pattern portion of the transcription of the audio communication and the phraseology pattern portion of the ground truth for the transcription; and
update the speech recognition vocabulary to include the nonstandard pattern based on the pattern-based performance metric.
16. A computing device comprising:
at least one computer-readable storage medium to store computer-executable instructions; and
at least one processor, coupled to the at least one computer-readable storage medium, to execute the computer-executable instructions to:
analyze a transcription of an audio communication with respect to a vehicle to characterize a nonstandard pattern within the transcription of the audio communication;
obtain a ground truth for the transcription of the audio communication;
determine one or more performance metrics associated with the nonstandard pattern within the transcription based on a relationship between the transcription of the audio communication and the ground truth for the transcription;
update a speech recognition vocabulary to include the nonstandard pattern based at least in part on the one or more performance metrics, resulting in an updated speech recognition vocabulary; and
determine an updated speech recognition model for the vehicle using the updated speech recognition vocabulary and the audio communication.
17. The computing device of claim 16 , wherein the computer-executable instructions cause the at least one processor to push the updated speech recognition model to the vehicle over a communications network.
18. The computing device of claim 16 , wherein the computer-executable instructions cause the at least one processor to analyze a second transcription of a subsequent audio communication with respect to the vehicle to detect the nonstandard pattern within the second transcription of the subsequent audio communication using at least one of the updated speech recognition vocabulary and the updated speech recognition model.
19. The computing device of claim 16 , wherein the computer-executable instructions cause the at least one processor to:
identify a phraseology pattern portion associated with the nonstandard pattern within the transcription of the audio communication; and
determine a phraseology pattern performance metric based on a relationship between the phraseology pattern portion of the transcription of the audio communication and a second phraseology pattern portion of the ground truth.
20. The computing device of claim 16 , wherein the computer-executable instructions cause the at least one processor to:
determine a pattern-based performance metric associated with the nonstandard pattern within the transcription based on a relationship between a phraseology pattern portion of the transcription of the audio communication and the phraseology pattern portion of the ground truth for the transcription; and
update the speech recognition vocabulary based on the pattern-based performance metric.