반응형
[Paper 리뷰] EmoMix: Emotion Mixing via Diffusion Models for Emotional Speech Synthesis
EmoMix: Emotion Mixing via Diffusion Models for Emotional Speech SynthesisEmotional Text-to-Speech는 여전히 intensity control 측면에서 한계가 있음EmoMixEmotion embedding을 추출하기 위해 pre-trained Speech Emotion Recognition model을 활용Run-time 시 diffusion model을 기반으로 mixed emotion synthesis를 수행논문 (INTERSPEECH 2023) : Paper Link1. IntroductionGenerSpeech와 같은 emotional Text-to-Speech (TTS) model은 reference-based style..
Paper/TTS
2025. 7. 4. 17:00
반응형