a-DCF: an architecture agnostic metric with application to spoofing-robust speaker verification

TL;DR

The paper addresses the challenge of fairly evaluating spoofing-robust ASV systems across diverse architectures. It proposes the architecture-agnostic detection cost function (a-DCF), a Bayes-risk based, multi-class extension of the NIST DCF that explicitly models target, non-target, and spoof priors with corresponding costs, and normalizes the score to an interpretable range. The approach generalizes existing metrics and accommodates non-tandem architectures, unlike t-DCF, while remaining applicable to various system types and scoring outputs. The study demonstrates the a-DCF on the ASVspoof 2019 LA dataset with cascade, jointly optimized, and single-model systems, illustrating its practical flexibility and interpretability for benchmarking spoofing-robust ASV. Overall, a-DCF enables architecture-agnostic, principled evaluation with explicit priors, potentially extending to other biometric domains.

Abstract

Spoofing detection is today a mainstream research topic. Standard metrics can be applied to evaluate the performance of isolated spoofing detection solutions and others have been proposed to support their evaluation when they are combined with speaker detection. These either have well-known deficiencies or restrict the architectural approach to combine speaker and spoof detectors. In this paper, we propose an architecture-agnostic detection cost function (a-DCF). A generalisation of the original DCF used widely for the assessment of automatic speaker verification (ASV), the a-DCF is designed for the evaluation of spoofing-robust ASV. Like the DCF, the a-DCF reflects the cost of decisions in a Bayes risk sense, with explicitly defined class priors and detection cost model. We demonstrate the merit of the a-DCF through the benchmarking evaluation of architecturally-heterogeneous spoofing-robust ASV solutions.