How Do Neural Spoofing Countermeasures Detect Partially Spoofed Audio?
This work addresses the interpretability gap in audio spoofing detection for security applications, though it is incremental as it builds on existing countermeasure methods.
The paper tackled the problem of understanding how neural spoofing countermeasures detect partially spoofed audio, finding that these models prioritize artifacts in transition regions rather than pattern differences, and introduced interpretability tools like Grad-CAM and a quantitative metric to analyze this.
Partially manipulating a sentence can greatly change its meaning. Recent work shows that countermeasures (CMs) trained on partially spoofed audio can effectively detect such spoofing. However, the current understanding of the decision-making process of CMs is limited. We utilize Grad-CAM and introduce a quantitative analysis metric to interpret CMs' decisions. We find that CMs prioritize the artifacts of transition regions created when concatenating bona fide and spoofed audio. This focus differs from that of CMs trained on fully spoofed audio, which concentrate on the pattern differences between bona fide and spoofed parts. Our further investigation explains the varying nature of CMs' focus while making correct or incorrect predictions. These insights provide a basis for the design of CM models and the creation of datasets. Moreover, this work lays a foundation of interpretability in the field of partial spoofed audio detection that has not been well explored previously.