--- license: mit task_categories: - text-to-speech language: - vi tags: - zonos2 - voice-cloning - vietnamese - colab - srt pretty_name: ZONOS2 Vietnamese Voices + Colab --- # ZONOS2 VI — giọng Việt + Colab (L4) Dataset phụ trợ cho [Zyphra/ZONOS2](https://huggingface.co/Zyphra/ZONOS2): - 6 giọng tham chiếu tiếng Việt (audio + transcript) - 2 notebook Colab (clone giọng + đọc SRT), runtime **GPU L4** > Model weights ZONOS2 **không** nằm trong repo này — notebook tải từ `Zyphra/ZONOS2`. ## Cấu trúc ``` voices// ref.wav | ref.mp3 ref_text.txt colab/ ZONOS2_VI_Clone_Colab.ipynb ZONOS2_VI_SRT_Colab.ipynb ``` ## Giọng | slug | file | |------|------| | ban_mai | Ban_Mai.mp3 | | lan_trinh | Lan_Trinh.wav | | ngan_ha | Ngan_Ha.wav | | ngoc_huyen | Ngoc_Huyen.mp3 | | thao_trinh | Thao_Trinh.wav | | tuong_vy | Tuong_Vy.wav | ## Colab 1. Runtime → **L4** (T4 không chạy ZONOS2) 2. Mở notebook Clone hoặc SRT 3. Đặt `HF_REPO = "STBack23/zonos2-vi"` (đã sẵn trong notebook) 4. Clone tạo `speaker.npy` trên Drive; SRT dùng profile đó ## License - Code / notebooks trong dataset: MIT (theo ZONOS2 inference) - Model weights `Zyphra/ZONOS2`: Apache-2.0 - Audio mẫu: dùng nội bộ / theo quyền sở hữu của bạn khi clone giọng mới