Multi-Channel Replay Speech Detection using Acoustic Maps
This addresses security vulnerabilities in voice assistants, though it is incremental as it builds on existing beamforming techniques.
The paper tackled replay attack detection in speaker verification systems by proposing acoustic maps as a spatial feature representation, achieving competitive performance on the ReMASC dataset with a lightweight CNN of about 6k parameters.
Replay attacks remain a critical vulnerability for automatic speaker verification systems, particularly in real-time voice assistant applications. In this work, we propose acoustic maps as a novel spatial feature representation for replay speech detection from multi-channel recordings. Derived from classical beamforming over discrete azimuth and elevation grids, acoustic maps encode directional energy distributions that reflect physical differences between human speech radiation and loudspeaker-based replay. A lightweight convolutional neural network is designed to operate on this representation, achieving competitive performance on the ReMASC dataset with approximately 6k trainable parameters. Experimental results show that acoustic maps provide a compact and physically interpretable feature space for replay attack detection across different devices and acoustic environments.