Extrapolating false alarm rates in automatic speaker verification
This provides a tool for ASV vendors and corpus providers to predict system reliability in adversarial scenarios, though it is incremental as it builds on existing score-space analysis.
The paper tackles the problem of extrapolating false alarm rates in automatic speaker verification by developing generative models that simulate worst-case adversarial impostors, enabling performance estimation for large speaker populations without new data collection.
Automatic speaker verification (ASV) vendors and corpus providers would both benefit from tools to reliably extrapolate performance metrics for large speaker populations without collecting new speakers. We address false alarm rate extrapolation under a worst-case model whereby an adversary identifies the closest impostor for a given target speaker from a large population. Our models are generative and allow sampling new speakers. The models are formulated in the ASV detection score space to facilitate analysis of arbitrary ASV systems.