# Open-Set Source Tracing as Compositional Factors via Structured Prototypes

Source: [Open-Set Source Tracing as Compositional Factors via Structured Prototypes](<https://www.sota2.com/research/paper/6a4c87e385dba69fef8b23d0>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-07-08T05:00:04.942Z

Authors: Santiago Rubio, Antonio Almud\\'evar, Antonio Miguel, Eduardo Lleida, Alfonso Ortega

Published: 2026-07-03T09:23:36.000Z

## Abstract

Recent research expands beyond binary anti-spoofing with the emergence of Source Tracing, the task of identifying the specific generative origins of synthetic speech. However, current research often equates a "source" with its generative architecture. We propose redefining a source as a compositional tuple of Architecture, Training Data, and other training factors affecting the generated speech. We propose a framework using Structured Orthonormal Prototypes to minimize class overlap and intra-class variance. Our Subspace Partitioning strategy splits the embedding into architecture and data subspaces, while a residual subspace captures stochastic variability, enabling "compositional generalization" for novel factor combinations. This approach improves performance for partially seen sources and maintains robustness in fully open-set scenarios. MLAAD evaluations for Few-Shot open-set Identification show our approach significantly outperforms angular-margin baselines.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Global Source Attribution on MLAAD Few-Shot Open-Set (test)](<https://www.sota2.com/research/sota/global-source-attribution-on-mlaad-few-shot-open-set-test.md>): Global Source Attribution; MLAAD Few-Shot Open-Set (test); F1-Macro (All vs All): 84.52.

- [Global Source Attribution on MLAAD Closed-Set IID (test)](<https://www.sota2.com/research/sota/global-source-attribution-on-mlaad-closed-set-iid-test.md>): Global Source Attribution; MLAAD Closed-Set IID (test); F1-Macro (Seen IID): 95.44.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2607.03134>)

[PDF](<https://arxiv.org/pdf/2607.03134.pdf>)