Federated Learning Framework for Privacy-Preserving Voice Analytics in Smart TV Ecosystems
International Journal of Computational and Experimental Science and Engineering · 30.12.2025 · 11 (4)
DOI: 10.22399/ijcesen.5119 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computational and Experimental Science and Engineering, online erschienen am 30.12.2025. Verfasst von Saraschandra Arveti und 2 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Voice assistants are increasingly being embedded within smart TVs, which is enabling consumers to interact with these devices more seamlessly and easily. One of the main ways that voice-based services service a user is by collecting and sending voice data via audio recordings to centralized cloud servers for analysis. While users may benefit from the information generated by such services, they will also face significant risks to their privacy and data security. Because most of the existing methods for machine learning rely solely on centralized machine learning data sets, the chances that sensitive private data will be improperly disclosed increases substantially. To mitigate these problems, the research described herein describes a federated framework for performing voice analytics in a privacy-preserving manner within the smart television ecosystem. The proposed federated framework will permit multiple smart televisions to collaboratively develop and train voice recognition models and perform analytics tasks, without transmitting any of the users' raw voice recordings to a single central server. Instead of transmitting raw recordings, the users' smart televisions will develop and train models locally using their own voice recordings. The smart televisions then transmit only encrypted updates regarding their models to a global aggregation server that will combine or aggregate them to create an overall or universal model. Implementing a federated learning approach significantly decreases the risk of an improper disclosure of personal information, while providing a voice recognition and analytics model that has comparable accuracy and efficiency as a central model. Additionally, the federated framework will implement secure protocols for the aggregation and communication of voice data to increase the security of the voice data and its robustness to possible attacks.
Zitieren
APA
Arveti, S., Hadkar, A., & Nutalapati, M. T. (2025). Federated Learning Framework for Privacy-Preserving Voice Analytics in Smart TV Ecosystems. International Journal of Computational and Experimental Science and Engineering, 11(4). https://doi.org/10.22399/ijcesen.5119
Vancouver
Arveti S, Hadkar A, Nutalapati MT. Federated Learning Framework for Privacy-Preserving Voice Analytics in Smart TV Ecosystems. International Journal of Computational and Experimental Science and Engineering. 2025;11(4). doi:10.22399/ijcesen.5119
BibTeX
@article{Arveti2025,
author = {Arveti, Saraschandra and Hadkar, Anish and Nutalapati, Mani Teja},
title = {Federated Learning Framework for Privacy-Preserving Voice Analytics in Smart TV Ecosystems},
journal = {International Journal of Computational and Experimental Science and Engineering},
year = {2025},
volume = {11},
number = {4},
doi = {10.22399/ijcesen.5119}
}
EUVIMED