Saved in:
Bibliographic Details
Main Authors: Li, Aoduo, Lv, Haoran, Xu, Hongjian, Li, Shengmin, Qin, Sihao, Li, Zimeng, Pun, Chi Man, Chen, Xuhang
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.19055
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908989016506368
author Li, Aoduo
Lv, Haoran
Xu, Hongjian
Li, Shengmin
Qin, Sihao
Li, Zimeng
Pun, Chi Man
Chen, Xuhang
author_facet Li, Aoduo
Lv, Haoran
Xu, Hongjian
Li, Shengmin
Qin, Sihao
Li, Zimeng
Pun, Chi Man
Chen, Xuhang
contents High-fidelity character voice synthesis is a cornerstone of immersive multimedia applications, particularly for interacting with anime avatars and digital humans. However, existing systems struggle to maintain consistent persona traits across diverse emotional contexts. To bridge this gap, we present ATRIE, a unified framework utilizing a Persona-Prosody Dual-Track (P2-DT) architecture. Our system disentangles generation into a static Timbre Track (via Scalar Quantization) and a dynamic Prosody Track (via Hierarchical Flow-Matching), distilled from a 14B LLM teacher. This design enables robust identity preservation (Zero-Shot Speaker Verification EER: 0.04) and rich emotional expression. Evaluated on our extended AnimeTTS-Bench (50 characters), ATRIE achieves state-of-the-art performance in both generation and cross-modal retrieval (mAP: 0.75), establishing a new paradigm for persona-driven multimedia content creation.
format Preprint
id arxiv_https___arxiv_org_abs_2604_19055
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ATRIE: Adaptive Tuning for Robust Inference and Emotion in Persona-Driven Speech Synthesis
Li, Aoduo
Lv, Haoran
Xu, Hongjian
Li, Shengmin
Qin, Sihao
Li, Zimeng
Pun, Chi Man
Chen, Xuhang
Sound
High-fidelity character voice synthesis is a cornerstone of immersive multimedia applications, particularly for interacting with anime avatars and digital humans. However, existing systems struggle to maintain consistent persona traits across diverse emotional contexts. To bridge this gap, we present ATRIE, a unified framework utilizing a Persona-Prosody Dual-Track (P2-DT) architecture. Our system disentangles generation into a static Timbre Track (via Scalar Quantization) and a dynamic Prosody Track (via Hierarchical Flow-Matching), distilled from a 14B LLM teacher. This design enables robust identity preservation (Zero-Shot Speaker Verification EER: 0.04) and rich emotional expression. Evaluated on our extended AnimeTTS-Bench (50 characters), ATRIE achieves state-of-the-art performance in both generation and cross-modal retrieval (mAP: 0.75), establishing a new paradigm for persona-driven multimedia content creation.
title ATRIE: Adaptive Tuning for Robust Inference and Emotion in Persona-Driven Speech Synthesis
topic Sound
url https://arxiv.org/abs/2604.19055