REIMU: Efficient Heterogeneous Hierarchical Reasoning for SSL-Based Speech Deepfake Detection

Authors: Kwok-Ho Ng, Tingting Song, Bingwen Feng, Peiya Li

Published: 2026-08-01 20:32:46+00:00

AI Summary

This research, named REIMU, investigates the effectiveness of recurrent hierarchical reasoning for speech deepfake detection using self-supervised learning (SSL) representations. It systematically compares various backbone architectures, including heterogeneous hierarchical designs combining self-attention with linear attention. The study concludes that while recurrence and hierarchical decomposition alone do not guarantee improvements, heterogeneous operator assignment offers a competitive and more parameter-efficient configuration.

Abstract

The increasing realism of speech generated by text-to-speech and voice conversion systems poses growing challenges to media integrity and voice authentication. Self-supervised learning (SSL) has substantially advanced speech deepfake detection, where downstream backbones conventionally process SSL representations through a single forward pass. This work investigates the practical effectiveness of recurrent hierarchical reasoning for this task. We term this controlled study REIMU and systematically compare conventional single-pass backbones, weight-shared recurrence, homogeneous HRM, and heterogeneous HRM across four Base-scale SSL frontends. We further examine heterogeneous high- and low-level modules that combine self-attention with linear attention. Experiments on the ASVspoof 2019 and 2021 evaluation sets show that recurrence and hierarchical decomposition do not inherently improve detection, whereas heterogeneous operator assignment provides a more competitive configuration. Notably, the heterogeneous design remains competitive while using 10.8\\% fewer downstream parameters than the matched baseline, demonstrating its potential for parameter-efficient speech deepfake detection.