ARCHITECTURAL APPROACHES TO BUILDING SCALABLE VOICE IDENTIFICATION SYSTEMS AND THEIR IMPACT ON IT PROJECT ECONOMISC

Abstract

The article examines architectural approaches to building scalable voice identification systems and their impact on the economics of IT projects in the field of software engineering and project management. The relevance of the topic is driven by the rapid growth of the voice services market and the need for businesses to simultaneously ensure high accuracy of biometric authentication, compliance with SLA requirements for latency and availability, as well as control of the total cost of ownership (TCO). The theoretical foundation of the study includes modern research on voice biometrics systems, microservice and cloud-native architectures, as well as cost management approaches for CAPEX/OPEX in cloud IT solutions. The aim of the article is to develop a conceptual model for correlating architectural decisions (monolithic vs. microservice architecture, on-premise vs. cloud deployment, use of modern ML frameworks) with key economic indicators of an IT project (CAPEX, OPEX, TCO, scaling-related savings, SLA non-compliance risks). The research methodology is based on a systematic analysis of scientific sources, comparative architectural analysis, formalization of economic dependencies, and constructing a generalized cost model for various deployment scenarios of voice identification systems. The obtained results include: a structural model of a typical voice identification system; an analytical comparison of monolithic and microservice architectures in terms of their impact on CAPEX/OPEX; an integrated economic model for evaluating TCO across four deployment options: monolithic on-premise, microservices on-premise, monolithic in the cloud (IaaS), and cloud-native/serverless architecture. It has been shown that transitioning to a microservice cloud-native architecture reduces the share of CAPEX and shifts costs toward a more controllable OPEX structure.