Instructions to use danhcao/dkvideo-zonos3-runtime with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Zonos
How to use danhcao/dkvideo-zonos3-runtime with Zonos:
# pip install git+https://github.com/Zyphra/Zonos.git import torchaudio from zonos.model import Zonos from zonos.conditioning import make_cond_dict model = Zonos.from_pretrained("danhcao/dkvideo-zonos3-runtime", device="cuda") wav, sr = torchaudio.load("speaker.wav") # 5-10s reference clip speaker = model.make_speaker_embedding(wav, sr) cond = make_cond_dict(text="Hello, world!", speaker=speaker, language="en-us") codes = model.generate(model.prepare_conditioning(cond)) audio = model.autoencoder.decode(codes)[0].cpu() torchaudio.save("sample.wav", audio, model.autoencoder.sampling_rate) - Notebooks
- Google Colab
- Kaggle
zonos3-server for Windows (x64, CUDA 13)
A Windows build of zonos3-server, the ggml server of
Zyphra/ZONOS3-Mini (servers/ggml), built unmodified from commit
96d9e0ec1e39083ed336ac78f24e3dbc93721c0f with MSVC 14.50, CMake + Ninja and CUDA 13.4
(-DGGML_CUDA=ON, CUDA architectures 75;80;86;89;120, AVX2).
| File | Contents |
|---|---|
zonos3-server-96d9e0ec-windows-x64-cuda13.zip |
zonos3-server.exe, the ggml DLLs, the Microsoft Visual C++ runtime DLLs, LICENSE |
zonos3-cudart-windows-x64-cuda13.4.zip |
cudart64_13.dll, cublas64_13.dll, cublasLt64_13.dll for machines without a CUDA 13 Toolkit |
The model weights are not here: get them from
Zyphra/ZONOS3-Mini (gguf/).
Licenses
zonos3-serverand ggml: MIT (seeLICENSEin the zip).- The CUDA runtime DLLs are NVIDIA redistributables under the NVIDIA CUDA Toolkit EULA.
- The Visual C++ runtime DLLs are Microsoft redistributables.