AudioNoisePrints: Model-free audio watermarking using spatial correlation in flow matching TTS
On the map: Diffusion models → Robustness, privacy and adversarial aspects
Abstract (arXiv)
We present AudioNoisePrints, a training-free watermarking pipeline for flow matching and diffusion TTS models, which requires minimal extra computation during inference and does not require retraining the TTS model or reducing the generation quality. We exploited the fact that there are strong correlations between the initial Gaussian noises and the generated outputs in diffusion and flow matching models, such that a simple cosine correlation between the initial noise and the generated output can be used to perform watermaking. Moreover, we train a lightweight detector on top for more aggressive augmentations. Our method outperforms AudioSeal, a strong baseline for audio watermarking under strong augmentations. We experimented on F5TTS and other TTS and vocoder models, and concluded that they all exhibit similar spatial correlation properties, suggesting our watermarking scheme can be used for more flow-matching TTS models and even vocoders in the future.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Robustness, privacy and adversarial aspects, one of 14 directions in Diffusion models. Builds on 20 of its 24 references that are in the map. First posted on arXiv 23 Aug 2026.
Builds on
20 of 24 references.
Closest ancestors · 2
Same direction, newest first
- 2025NoisePrints: Distortion-Free Watermarks for Authorship in Private Diffusion Models15 Oct
- 2024GROOT: Generating Robust Watermark for Diffusion-Model-Based Audio Synthesis15 Jul
Other references · 18
- 2020Denoising Diffusion Probabilistic Modelscited by 21,076
- 2022Scalable Diffusion Models with Transformerscited by 6,054
- 2022Flow Matching for Generative Modelingcited by 5,702
- 2020DiffWave: A Versatile Diffusion Model for Audio Synthesiscited by 1,354
- 2019LibriTTS: A Corpus Derived from LibriSpeech for Text-to-Speechcited by 649
- 2021VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation Learning, Semi-Supervised Learning and Interpretationcited by 463
- 2024F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matchingcited by 361
- 2023Vocos: Closing the gap between time-domain and Fourier-based neural vocoders for high-quality audio synthesiscited by 153
- 2024E2 TTS: Embarrassingly Easy Fully Non-Autoregressive Zero-Shot TTScited by 152
- 2024Proactive Detection of Voice Cloning with Localized Watermarkingcited by 78
- 2025The Principles of Diffusion Modelscited by 76
- 2023WavMark: Watermarking for Audio Generationcited by 50
6 more
- 2023Detecting Voice Cloning Attacks via Timbre Watermarkingcited by 44
- 2024Latent Watermarking of Audio Generative Modelscited by 16
- 2025A Comprehensive Real-World Assessment of Audio Watermarking Algorithms: Will They Survive Neural Codecs?cited by 13
- 2025WaveFM: A High-Fidelity and Efficient Vocoder Based on Flow Matchingcited by 7
- 2025Yours or Mine? Overwriting Attacks Against Neural Audio Watermarkingcited by 5
- 2025Traceable TTS: Toward Watermark-Free TTS with Strong Traceabilitycited by 1
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Multi-bit Audio Watermarking
- 2023Invisible Watermarking for Audio Generation Diffusion Models
- 2026Hidden in Plain Tokens: Simply Robust, Gradient-Free Watermark for Synthetic Audio
- 2026mAVE: A Watermark for Joint Audio-Visual Generation Models
- 2024SilentCipher: Deep Audio Watermarking
- 2024AudioMarkBench: Benchmarking Robustness of Audio Watermarking
- 2025XAttnMark: Learning Robust Audio Watermarking with Cross-Attention
- 2025HarmonicAttack: An Adaptive Cross-Domain Audio Watermark Removal