CosyEdit2: Speech-Editing-Oriented Reinforcement Learning Unlocks Better Zero-Shot TTS
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911719575519232 |
|---|---|
| author | Chen, Junyang Jia, Yuhang Wang, Hui Zhou, Jiaming Gan, Yongchang Qin, Yong |
| author_facet | Chen, Junyang Jia, Yuhang Wang, Hui Zhou, Jiaming Gan, Yongchang Qin, Yong |
| contents | Speech editing and zero-shot Text-to-Speech (TTS) share a similar generative foundation conditioned on speech prompts, yet speech editing demands far stricter local acoustic consistency with surrounding unedited content. While prior work has shown that Supervised Fine-Tuning (SFT) enables TTS models to acquire functional editing capability, this approach remains fundamentally bottlenecked by imperfect paired editing data and coarse-grained optimization signals. To address these limitations, we propose CosyEdit2, a speech editing model built on a two-stage post-training framework that progresses from supervised editing initialization to editing-oriented Group Relative Policy Optimization (GRPO) over target-speech-free data. Extensive experiments demonstrate that CosyEdit2 not only substantially advances speech editing performance, but also unlocks better zero-shot TTS capability, revealing a deeper mutual relationship between the two tasks. Audio samples are available at https://cjy1018.github.io/CosyEdit2. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_25930 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CosyEdit2: Speech-Editing-Oriented Reinforcement Learning Unlocks Better Zero-Shot TTS Chen, Junyang Jia, Yuhang Wang, Hui Zhou, Jiaming Gan, Yongchang Qin, Yong Sound Speech editing and zero-shot Text-to-Speech (TTS) share a similar generative foundation conditioned on speech prompts, yet speech editing demands far stricter local acoustic consistency with surrounding unedited content. While prior work has shown that Supervised Fine-Tuning (SFT) enables TTS models to acquire functional editing capability, this approach remains fundamentally bottlenecked by imperfect paired editing data and coarse-grained optimization signals. To address these limitations, we propose CosyEdit2, a speech editing model built on a two-stage post-training framework that progresses from supervised editing initialization to editing-oriented Group Relative Policy Optimization (GRPO) over target-speech-free data. Extensive experiments demonstrate that CosyEdit2 not only substantially advances speech editing performance, but also unlocks better zero-shot TTS capability, revealing a deeper mutual relationship between the two tasks. Audio samples are available at https://cjy1018.github.io/CosyEdit2. |
| title | CosyEdit2: Speech-Editing-Oriented Reinforcement Learning Unlocks Better Zero-Shot TTS |
| topic | Sound |
| url | https://arxiv.org/abs/2605.25930 |