DevoX: Deep Voice Detection MLOps Framework for Crime Prevention in the Core-Edge Cloud

Yuri Seo, Hyeon-ki Jo, Seol Roh, Teh-Jen Sun, Seung-Woo Jeong, HakHo Kim, Thien-Thu Ngo, Eui-Nam Huh

Abstract: With the increase in criminal cases using deep voice, research related to deep voice detection is on the rise. However, the previous studies of deep voice detection are still insufficient when applied to actual crime prevention services. To address this issue, this paper proposes a new framework called "DevoX" that can apply deep voice detection to actual services. This framework continually trains on voice data from end devices (such as smartphones) for new types of deep voice crimes. Also, DevoX provides low-latency services through the core-edge cloud architecture. Edge cloud sends extracted features through encoding instead of call voice to prevent privacy leaks. It is possible to apply deep voice detection technology to actual services. Therefore, our study can be effectively utilized in preventing phishing based on deep voice.

External IDs:dblp:conf/icoin/SeoJRSJKNH24