Saved in:
Bibliographic Details
Main Authors: Lu, Renjie, Zhang, Xulong, Qu, Xiaoyang, Wang, Jianzong, Wang, Shangfei
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.22501
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908799107858432
author Lu, Renjie
Zhang, Xulong
Qu, Xiaoyang
Wang, Jianzong
Wang, Shangfei
author_facet Lu, Renjie
Zhang, Xulong
Qu, Xiaoyang
Wang, Jianzong
Wang, Shangfei
contents Synthesizing personalized talking faces that uphold and highlight a speaker's unique style while maintaining lip-sync accuracy remains a significant challenge. A primary limitation of existing approaches is the intrinsic confounding of speaker-specific talking style and semantic content within facial motions, which prevents the faithful transfer of a speaker's unique persona to arbitrary speech. In this paper, we propose MirrorTalk, a generative framework based on a conditional diffusion model, combined with a Semantically-Disentangled Style Encoder (SDSE) that can distill pure style representations from a brief reference video. To effectively utilize this representation, we further introduce a hierarchical modulation strategy within the diffusion process. This mechanism guides the synthesis by dynamically balancing the contributions of audio and style features across distinct facial regions, ensuring both precise lip-sync accuracy and expressive full-face dynamics. Extensive experiments demonstrate that MirrorTalk achieves significant improvements over state-of-the-art methods in terms of lip-sync accuracy and personalization preservation.
format Preprint
id arxiv_https___arxiv_org_abs_2601_22501
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MIRRORTALK: Forging Personalized Avatars Via Disentangled Style and Hierarchical Motion Control
Lu, Renjie
Zhang, Xulong
Qu, Xiaoyang
Wang, Jianzong
Wang, Shangfei
Computer Vision and Pattern Recognition
Sound
Synthesizing personalized talking faces that uphold and highlight a speaker's unique style while maintaining lip-sync accuracy remains a significant challenge. A primary limitation of existing approaches is the intrinsic confounding of speaker-specific talking style and semantic content within facial motions, which prevents the faithful transfer of a speaker's unique persona to arbitrary speech. In this paper, we propose MirrorTalk, a generative framework based on a conditional diffusion model, combined with a Semantically-Disentangled Style Encoder (SDSE) that can distill pure style representations from a brief reference video. To effectively utilize this representation, we further introduce a hierarchical modulation strategy within the diffusion process. This mechanism guides the synthesis by dynamically balancing the contributions of audio and style features across distinct facial regions, ensuring both precise lip-sync accuracy and expressive full-face dynamics. Extensive experiments demonstrate that MirrorTalk achieves significant improvements over state-of-the-art methods in terms of lip-sync accuracy and personalization preservation.
title MIRRORTALK: Forging Personalized Avatars Via Disentangled Style and Hierarchical Motion Control
topic Computer Vision and Pattern Recognition
Sound
url https://arxiv.org/abs/2601.22501