반응형
[Paper 리뷰] EMORL-TTS: Reinforcement Learning for Fine-Grained Emotion Control in LLM-based TTS
EMORL-TTS: Reinforcement Learning for Fine-Grained Emotion Control in LLM-based TTSLarge Language Model-based Text-to-Speech model은 fine-grained emotional control 측면에서 한계가 있음EMORL-TTSVAD space의 global intensity control과 local emphasis regulation을 unify 함특히 emotion category, intensity, emphasis의 task-specific reward를 통해 guide 되는 reinforcement learning과 supervised fine-tuning을 combine 함논문 (ICASSP ..
Paper/TTS
2026. 3. 5. 12:57
반응형