The Binding Effect: Analysis of How Multi-Dimensional Cues Form Gender Bias in Instruction TTS
Kuan-Yu, Yi-Cheng, Po-Chung, Huang-Cheng, Chih-Fan, Jeng-Lin, Hung-yi, Jian-Jiun · 2026
Read The Binding Effect: Analysis of How Multi-Dimensional Cues Form Gender Bias in Instruction TTS by Kuan-Yu, Yi-Cheng, Po-Chung, Huang-Cheng, Chih-Fan, Jeng-Lin, Hung-yi, Jian-Jiun (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-20743.txt.
Description
Chen and co-authors study gender bias in instruction text to speech systems. Most past tests vary one cue at a time. Real prompts mix social status, career cues, and persona traits at once. The team models prompts as blends of these three cue types. Strong interaction effects showed up in open source models. Bias tied back to text encoder priors and skewed training data. Plain diversity prompts failed to break these patterns. They argue bias checks must test cue blends, not single cues.
Catalog record
| CREATOR | Kuan-Yu, Yi-Cheng, Po-Chung, Huang-Cheng, Chih-Fan, Jeng-Lin, Hung-yi, Jian-Jiun |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25549 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.20743