Efficient Speech Enhancement via Embeddings from Pre-trained Generative Audioencoders

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Sun, Xingwei, Dinkel, Heinrich, Niu, Yadong, Wang, Linzhang, Zhang, Junbo, Luan, Jian
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909648346415104
author Sun, Xingwei
Dinkel, Heinrich
Niu, Yadong
Wang, Linzhang
Zhang, Junbo
Luan, Jian
author_facet Sun, Xingwei
Dinkel, Heinrich
Niu, Yadong
Wang, Linzhang
Zhang, Junbo
Luan, Jian
contents Recent research has delved into speech enhancement (SE) approaches that leverage audio embeddings from pre-trained models, diverging from time-frequency masking or signal prediction techniques. This paper introduces an efficient and extensible SE method. Our approach involves initially extracting audio embeddings from noisy speech using a pre-trained audioencoder, which are then denoised by a compact encoder network. Subsequently, a vocoder synthesizes the clean speech from denoised embeddings. An ablation study substantiates the parameter efficiency of the denoise encoder with a pre-trained audioencoder and vocoder. Experimental results on both speech enhancement and speaker fidelity demonstrate that our generative audioencoder-based SE system outperforms models utilizing discriminative audioencoders. Furthermore, subjective listening tests validate that our proposed system surpasses an existing state-of-the-art SE model in terms of perceptual quality.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11514
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Speech Enhancement via Embeddings from Pre-trained Generative Audioencoders
Sun, Xingwei
Dinkel, Heinrich
Niu, Yadong
Wang, Linzhang
Zhang, Junbo
Luan, Jian
Audio and Speech Processing
Sound
Recent research has delved into speech enhancement (SE) approaches that leverage audio embeddings from pre-trained models, diverging from time-frequency masking or signal prediction techniques. This paper introduces an efficient and extensible SE method. Our approach involves initially extracting audio embeddings from noisy speech using a pre-trained audioencoder, which are then denoised by a compact encoder network. Subsequently, a vocoder synthesizes the clean speech from denoised embeddings. An ablation study substantiates the parameter efficiency of the denoise encoder with a pre-trained audioencoder and vocoder. Experimental results on both speech enhancement and speaker fidelity demonstrate that our generative audioencoder-based SE system outperforms models utilizing discriminative audioencoders. Furthermore, subjective listening tests validate that our proposed system surpasses an existing state-of-the-art SE model in terms of perceptual quality.
title Efficient Speech Enhancement via Embeddings from Pre-trained Generative Audioencoders
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2506.11514