[
    {
        "id": "osp-16075",
        "type": "article-journal",
        "title": "Task-Aware Joint Pruning and Distillation for Efficient Audio Deepfake Detection",
        "author": [
            {
                "family": "He",
                "given": "Miao"
            },
            {
                "family": "Cheng",
                "given": "Peng"
            },
            {
                "family": "Ba",
                "given": "Zhongjie"
            },
            {
                "family": "Wen",
                "given": "Qing"
            },
            {
                "family": "Lu",
                "given": "Li"
            },
            {
                "family": "Yang",
                "given": "Xin"
            },
            {
                "family": "Ren",
                "given": "Kui"
            }
        ],
        "URL": "https://omanscience.com/ar/articles/task-aware-joint-pruning-and-distillation-for-efficient-audio-deepfake-detection",
        "language": "en",
        "issued": {
            "date-parts": [
                [
                    2026
                ]
            ]
        },
        "DOI": "10.21437/interspeech.2026-1766",
        "abstract": "Advances in speech synthesis have made deepfake speeches increasingly convincing, posing growing threats to security. While self-supervised learning (SSL) based detectors achieve state-of-the-art performance, their computational demands (typically 300M+ parameters) prevent deployment on resource-constrained devices. Existing compression methods, designed mainly for content-centric tasks, struggle to maintain competitive performance when directly adapted to deepfake detection. We propose a Task-Aware Joint Pruning and Distillation framework that combines cross-domain knowledge distillation with movement-guided structured pruning to transfer forgery-discriminative knowledge and preserve critical structures under aggressive compression. Our framework reduces the model to 31.9M parameters with 6.3$\\times$ FLOPs reduction, with an average performance drop of only 1.30\\% across multiple datasets compared to the uncompressed baseline, demonstrating strong potential for on-device deployment."
    }
]