Publication Details
Detecting English Speech in the Air Traffic Control Voice Communication
Kesiraju Santosh, Ph.D. (DCGM)
Novotný Ondřej, Ing., Ph.D.
Kocour Martin, Ing. (DCGM)
Veselý Karel, Ing., Ph.D. (DCGM)
Černocký Jan, prof. Dr. Ing. (DCGM)
speech recognition, language detection, x-vectorextractor, acoustic model, air-traffic communication, data collection, text embeddings, Bayesian methods
Developing in-cockpit voice enabled applications require a realworlddataset with labels and annotations. We launched a communityplatform for collecting the Air-Traffic Control (ATC)speech, world-wide in the ATCO2 project. Filtering out non-English speech is one of the main components in the data processingpipeline. The proposed English Language Detection(ELD) system is based on the embeddings from Bayesian subspacemultinomial model. It is trained on the word confusionnetwork from an ASR system. It is robust, easy to train, andlight weighted. We achieved 0:0439 equal-error-rate (EER),a 50% relative reduction as compared to the state-of-the-artacoustic ELD system based on x-vectors, in the in-domain scenario.Further, we achieved an EER of 0:1352, a 33% relativereduction as compared to the acoustic ELD, in the unseen language(out-of-domain) condition. We plan to publish the evaluationdataset from the ATCO2 project.
@inproceedings{BUT175844,
author="Igor {Szőke} and Santosh {Kesiraju} and Ondřej {Novotný} and Martin {Kocour} and Karel {Veselý} and Jan {Černocký}",
title="Detecting English Speech in the Air Traffic Control Voice Communication",
booktitle="Proceedings Interspeech 2021",
year="2021",
journal="Proceedings of Interspeech",
volume="2021",
number="8",
pages="3286--3290",
publisher="International Speech Communication Association",
address="Brno",
doi="10.21437/Interspeech.2021-1033",
issn="1990-9772",
url="https://www.isca-speech.org/archive/interspeech_2021/szoke21_interspeech.html"
}