elib (German Aerospace Center) · 2017 · 11 citations · 8 references
ComputingAutonomous SystemsComputer ScienceArtificial IntelligenceAutomationMachine LearningHuman-machine InterfacesPattern RecognitionSpeech RecognitionSupervised LearningAutomatic RecognitionCommandsCustomizationEnd-to-end Speech RecognitionSpeech Pattern RecognitionSpeech Recognition AccuracySpeech Recognition EvaluationSpeech Understanding
Recently, the project AcListant® related to automatic speech recognition has achieved command recognition error rates below 1.7% based on Assistant Based Speech Recognition (ABSR). One main issue to transfer ABSR from the laboratory to the ops-rooms is its costs of deployment. Currently each ABSR model must manually be adapted to the local environment due to e.g. different accents and models to predict possible controller commands. The Horizon 2020 funded project MALORCA (Ma-chine Learning of Speech Recognition Models for Controller As-sistance) proposes a general, cheap and effective solution to au-tomate this re-learning, adaptation and customization process to new environments, by taking advantage of the large amount of speech data available in the ATM world. This paper presents an algorithm which automatically learns a model to predict control-ler commands from recorded untranscribed controller utterances and the corresponding radar data. The trained model is validated against transcribed controller commands for Vienna and Prague approach. Command error rates are reduced from 4.1% to 0.9% for Prague approach and from 10.9% to 2.0% for Vienna ap-proach.
8
J. R. Quinlan · Machine Learning · 1986 · 14.8K citations · Full text
High level knowledge sources in usable speech recognition systems
S. L. Young, Alexander G. Hauptmann, Wayne Ward et al. · Communications of the ACM · 1989 · 115 citations · Full text
Reducing controller workload with automatic speech recognition
Hartmut Helmke, Oliver Ohneiser, Thorsten Mühlhausen et al. · 2016 · 49 citations