arXiv cs.CL· Joonyong Park, Jerry Li·· 14 小时前AI 评分27
当人设与台词冲突：诊断 Chain-of-Thought TTS 中的计划到合成失败
When the Persona Contradicts the Line: Diagnosing Plan-to-Synthesis Failures in Chain-of-Thought TTS
AI 导读
研究构建了日英双语人设—台词冲突基准，每项将台词与一致及冲突人设配对并预先标注期望的声学方向，用于追踪 Chain-of-Thought TTS 中意图在规划短语、规划器情感向量、QwenEmotion 向量和渲染波形四个环节的丢失情况。
正文 · 原文
Abstract:Chain-of-thought TTS systems rely on language-model planners to translate persona and context into vocal delivery, but it is unclear where this intent is lost when the requested affect conflicts with lexical sentiment. We introduce a bilingual Japanese--English persona--line conflict benchmark in which each item pairs one line with congruent and conflicting personas and pre-declares the expected acoustic direction. We probe this direction in the planner-generated delivery phrase, the planner-written emotion vector, the QwenEmotion-generated vector, and the rendered waveform. Planner outputs are usually direction-correct, while downstream agreement can be substantially lower. Bypassing QwenEmotion provides only modest gains over the full planner-controlled Base. Supervising QwenEmotion raises the direction rate to 75% in Japanese and 84% in English. A learned text-to-latent interface reaches the highest English rate of 89% and the highest Japanese arousal rate of 88%. These results show that learned repair of the plan-to-synthesis interface can recover counter-textual delivery, although its effect differs across languages and acoustic axes.
| Comments: | Submitted to ICASSP 2027 |
| Subjects: | Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS) |
| Cite as: | arXiv:2610.35008 [cs.CL] |
| (or arXiv:2610.35008v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35008 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Joonyong Park [view email]
[v1]
Mon, 26 Oct 2026 08:20:47 UTC (123 KB)
来源：arXiv cs.CL · arxiv.org