WaveNeXt 2: ConvNeXt-Based Fast Neural Vocoders With Residual Denoising and Sub-Modeling for GAN and Diffusion Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhou, Wangzixi, Okamoto, Takuma, Ohtani, Yamato, Sakti, Sakriani, Kawai, Hisashi
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910254380351488
author Zhou, Wangzixi
Okamoto, Takuma
Ohtani, Yamato
Sakti, Sakriani
Kawai, Hisashi
author_facet Zhou, Wangzixi
Okamoto, Takuma
Ohtani, Yamato
Sakti, Sakriani
Kawai, Hisashi
contents Most neural vocoders are limited to one type: either GAN or diffusion-based. While state-of-the-art models like Vocos and WaveNeXt use powerful ConvNeXt-based generators, they have only been used in GAN frameworks and have limited performance in multi-speaker settings. Moreover, diffusion models, despite training faster than GANs, have slow CPU inference. In this paper, we introduce WaveNeXt 2, a unified ConvNeXt-based framework compatible with both GAN and diffusion vocoders. Its core innovation is residual denoising and sub-modeling, where each sub-model progressively refines the waveform. Experimental results in the multi-speaker dataset demonstrate the effectiveness of our approach: (1) GAN-WaveNeXt 2 is much faster than HiFi-GAN and WaveFit, and (2) Diff-WaveNeXt 2 also delivers much faster inference and competitive synthesis quality compared with FastDiff with 4 steps. The Diff-WaveNeXt 2 is very training-efficient, training in only 32 hours, making it ideal for resource-constrained applications.
format Preprint
id arxiv_https___arxiv_org_abs_2605_25506
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle WaveNeXt 2: ConvNeXt-Based Fast Neural Vocoders With Residual Denoising and Sub-Modeling for GAN and Diffusion Models
Zhou, Wangzixi
Okamoto, Takuma
Ohtani, Yamato
Sakti, Sakriani
Kawai, Hisashi
Audio and Speech Processing
Most neural vocoders are limited to one type: either GAN or diffusion-based. While state-of-the-art models like Vocos and WaveNeXt use powerful ConvNeXt-based generators, they have only been used in GAN frameworks and have limited performance in multi-speaker settings. Moreover, diffusion models, despite training faster than GANs, have slow CPU inference. In this paper, we introduce WaveNeXt 2, a unified ConvNeXt-based framework compatible with both GAN and diffusion vocoders. Its core innovation is residual denoising and sub-modeling, where each sub-model progressively refines the waveform. Experimental results in the multi-speaker dataset demonstrate the effectiveness of our approach: (1) GAN-WaveNeXt 2 is much faster than HiFi-GAN and WaveFit, and (2) Diff-WaveNeXt 2 also delivers much faster inference and competitive synthesis quality compared with FastDiff with 4 steps. The Diff-WaveNeXt 2 is very training-efficient, training in only 32 hours, making it ideal for resource-constrained applications.
title WaveNeXt 2: ConvNeXt-Based Fast Neural Vocoders With Residual Denoising and Sub-Modeling for GAN and Diffusion Models
topic Audio and Speech Processing
url https://arxiv.org/abs/2605.25506