WMCodec: End-to-End Neural Speech Codec with Deep Watermarking for Authenticity Verification

2024-09-18

arXiv.org

AI Summary

To address the poor robustness, low capacity, and insufficient detection accuracy of speech watermarking caused by decoupled training of watermarking and speech codecs, this paper proposes the first end-to-end jointly optimized neural speech codec, enabling simultaneous high-fidelity compression reconstruction and robust watermark embedding/extraction. We innovatively design an iterative Attention Imprint Unit (AIU) that deeply fuses speech and watermark features; introduce a quantization-noise-aware watermarking mechanism; and optimize via a unified loss function. At 6 kbps bitrate and 16 bps watermark payload, the method achieves >99% watermark extraction accuracy under common attacks. It significantly outperforms AudioSeal+Encodec and TraceableSpeech in both watermark imperceptibility and extraction accuracy, thereby substantially enhancing authenticity assurance for speech content.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Speech
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

Recent advances in speech spoofing necessitate stronger verification mechanisms in neural speech codecs to ensure authenticity. Current methods embed numerical watermarks before compression and extract them from reconstructed speech for verification, but face limitations such as separate training processes for the watermark and codec, and insufficient cross-modal information integration, leading to reduced watermark imperceptibility, extraction accuracy, and capacity. To address these issues, we propose WMCodec, the first neural speech codec to jointly train compression-reconstruction and watermark embedding-extraction in an end-to-end manner, optimizing both imperceptibility and extractability of the watermark. Furthermore, We design an iterative Attention Imprint Unit (AIU) for deeper feature integration of watermark and speech, reducing the impact of quantization noise on the watermark. Experimental results show WMCodec outperforms AudioSeal with Encodec in most quality metrics for watermark imperceptibility and consistently exceeds both AudioSeal with Encodec and reinforced TraceableSpeech in extraction accuracy of watermark. At bandwidth of 6 kbps with a watermark capacity of 16 bps, WMCodec maintains over 99% extraction accuracy under common attacks, demonstrating strong robustness.

Problem

• Watermarking
• Audio Compression
• Authentication

Innovation

• Joint Compression and Watermarking
• AIU Mechanism
• Improved Robustness and Transparency