Enhanced Speech Detection Models for Air Travel Disease Risk

Auteurs-es

Mots-clés :

Speaker diarization, artificial intelligence, airport, aircraft cabin

Résumé

In noisy environments such as airport terminals or aircraft cabins, travelers tend to speak louder and to draw closer to their interlocutors, increasing the risk of respiratory particle dispersion and contagion. Accurate assessment of this risk is crucial amidst multiple overlapping speech sources. Utilizing advanced signal processing techniques, particularly artificial intelligence-based speaker diarization, this study focuses on accurately determining speaker identities, speech parameters, and sound levels without content disclosure. The paper describes tailored speaker diarization algorithms and presents validation and preliminary results. By analyzing speech features, the algorithms are used to calculate speech duration and sound pressure levels for each speaker and sentence, to assess viral contaminant spread. The article presents the implemented speaker diarization algorithms and the results for noisy, multi-participant environments, as part of a broader project effort aiming at proactively develop disease containment measures in air travel settings.

Fichiers supplémentaires

Publié-e

2024-05-12

Comment citer

1.
KONE TC, Ghinet S, Grewal A, Sayed Ahmed D. Enhanced Speech Detection Models for Air Travel Disease Risk. Canadian Acoustics [Internet]. 12 mai 2024 [cité 24 nov. 2024];52(1). Disponible à: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/4154

Articles les plus lus du,de la,des même-s auteur-e-s

1 2 3 4 > >>