Guardado en:
Detalles Bibliográficos
Autores principales: Yang, Gang, Lei, Yue, Tai, Wenxin, Wu, Jin, Chen, Jia, Zhong, Ting, Zhou, Fan
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2509.15952
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911167832653824
author Yang, Gang
Lei, Yue
Tai, Wenxin
Wu, Jin
Chen, Jia
Zhong, Ting
Zhou, Fan
author_facet Yang, Gang
Lei, Yue
Tai, Wenxin
Wu, Jin
Chen, Jia
Zhong, Ting
Zhou, Fan
contents Diffusion and flow matching (FM) models have achieved remarkable progress in speech enhancement (SE), yet their dependence on multi-step generation is computationally expensive and vulnerable to discretization errors. Recent advances in one-step generative modeling, particularly MeanFlow, provide a promising alternative by reformulating dynamics through average velocity fields. In this work, we present COSE, a one-step FM framework tailored for SE. To address the high training overhead of Jacobian-vector product (JVP) computations in MeanFlow, we introduce a velocity composition identity to compute average velocity efficiently, eliminating expensive computation while preserving theoretical consistency and achieving competitive enhancement quality. Extensive experiments on standard benchmarks show that COSE delivers up to 5x faster sampling and reduces training cost by 40%, all without compromising speech quality. Code is available at https://github.com/ICDM-UESTC/COSE.
format Preprint
id arxiv_https___arxiv_org_abs_2509_15952
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Compose Yourself: Average-Velocity Flow Matching for One-Step Speech Enhancement
Yang, Gang
Lei, Yue
Tai, Wenxin
Wu, Jin
Chen, Jia
Zhong, Ting
Zhou, Fan
Sound
Artificial Intelligence
Machine Learning
Audio and Speech Processing
Diffusion and flow matching (FM) models have achieved remarkable progress in speech enhancement (SE), yet their dependence on multi-step generation is computationally expensive and vulnerable to discretization errors. Recent advances in one-step generative modeling, particularly MeanFlow, provide a promising alternative by reformulating dynamics through average velocity fields. In this work, we present COSE, a one-step FM framework tailored for SE. To address the high training overhead of Jacobian-vector product (JVP) computations in MeanFlow, we introduce a velocity composition identity to compute average velocity efficiently, eliminating expensive computation while preserving theoretical consistency and achieving competitive enhancement quality. Extensive experiments on standard benchmarks show that COSE delivers up to 5x faster sampling and reduces training cost by 40%, all without compromising speech quality. Code is available at https://github.com/ICDM-UESTC/COSE.
title Compose Yourself: Average-Velocity Flow Matching for One-Step Speech Enhancement
topic Sound
Artificial Intelligence
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2509.15952