Mitigating data replication in text-to-audio generative diffusion models through anti-memorization guidance
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911404801392640 |
|---|---|
| author | Messina, Francisco Ronchini, Francesca Comanducci, Luca Bestagini, Paolo Antonacci, Fabio |
| author_facet | Messina, Francisco Ronchini, Francesca Comanducci, Luca Bestagini, Paolo Antonacci, Fabio |
| contents | A persistent challenge in generative audio models is data replication, where the model unintentionally generates parts of its training data during inference. In this work, we address this issue in text-to-audio diffusion models by exploring the use of anti-memorization strategies. We adopt Anti-Memorization Guidance (AMG), a technique that modifies the sampling process of pre-trained diffusion models to discourage memorization. Our study explores three types of guidance within AMG, each designed to reduce replication while preserving generation quality. We use Stable Audio Open as our backbone, leveraging its fully open-source architecture and training dataset. Our comprehensive experimental analysis suggests that AMG significantly mitigates memorization in diffusion-based text-to-audio generation without compromising audio fidelity or semantic alignment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_14934 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Mitigating data replication in text-to-audio generative diffusion models through anti-memorization guidance Messina, Francisco Ronchini, Francesca Comanducci, Luca Bestagini, Paolo Antonacci, Fabio Audio and Speech Processing Machine Learning Sound Signal Processing A persistent challenge in generative audio models is data replication, where the model unintentionally generates parts of its training data during inference. In this work, we address this issue in text-to-audio diffusion models by exploring the use of anti-memorization strategies. We adopt Anti-Memorization Guidance (AMG), a technique that modifies the sampling process of pre-trained diffusion models to discourage memorization. Our study explores three types of guidance within AMG, each designed to reduce replication while preserving generation quality. We use Stable Audio Open as our backbone, leveraging its fully open-source architecture and training dataset. Our comprehensive experimental analysis suggests that AMG significantly mitigates memorization in diffusion-based text-to-audio generation without compromising audio fidelity or semantic alignment. |
| title | Mitigating data replication in text-to-audio generative diffusion models through anti-memorization guidance |
| topic | Audio and Speech Processing Machine Learning Sound Signal Processing |
| url | https://arxiv.org/abs/2509.14934 |