TY  - RPRT
TI  - Data-efficient Targeted Token-level Preference Optimization for LLM-based Text-to-Speech
AU  - Rikuto Kotoge
AU  - Yuichi Sasaki
PY  - 2026
UR  - https://arxiv.org/abs/2510.05799
ID  - 2510.05799
ER  -