COCKTAIL
A tool to help air traffic controllers communicate
effectively in noisy environments
A tool to help air traffic controllers communicate
effectively in noisy environments
Communications with Operational Context and Knowledge for Target Audio Identification Learning (COCKTAIL)
Charles River Analytics is using novel natural language processing (NLP) techniques, including developing custom speech models for understanding, and automatically generating novel speech on the fly. COCKTIAL applies these techniques to help train air traffic controllers to communicate effectively in a noisy environment.
“When we improve the techniques for using speech processing technology to understand and generate communication for training applications, it will have wide applicability across the military because of how central training is to their work.”
![](https://b3069131.smushcdn.com/3069131/wp-content/uploads/2024/01/terry-patten_circle_headshot-1-200x200.png?lossy=1&strip=1&webp=1)
Dr. Terry Patten
Principal Scientist at Charles River Analytics and Principal Investigator on COCKTAIL
Charles River Analytics and partner OWT Global are developing techniques to tailor speech processing to specialized training applications, funded by the Naval Air Systems Command Naval Air Warfare Center Training Systems Division (NAWCTSD).
![](https://b3069131.smushcdn.com/3069131/wp-content/uploads/2024/03/COCKTAIL_SA-1024x508.jpg?lossy=1&strip=1&webp=1)
Speech recognition and generation are advanced in consumer electronics, such as virtual assistants. However, speech processing is still challenging in environments such as defense training programs with its specialized vocabulary and grammar. Commercial speech understanding systems have trouble with military language, which contains idiosyncratic vocabulary and grammar. Generating speech for training is also a challenge.
For example, air traffic control trainees use recorded speech to practice identifying important communication in a noisy background, but trainees quickly learn to recognize and ignore the recordings, which gives them an unfair advantage and makes the training less effective.
COCKTAIL will immediately support target audio identification training in the Navy, Air Force, and the Federal Aviation Administration. Furthermore, it could enhance Charles River’s KWYN™ suite of training solutions and commercial applications that require speech processing in training systems.
Contact us to learn more about COCKTAIL and our other adaptive training and artificial intelligence capabilities.
This material is based upon work supported by the Naval Air Systems Command (NAVAIR) under Contract No. N68335-23-C-0300. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of NAVAIR.