The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads model weights from Hugging Face (k2-fsa/OmniVoice) and supports installation via official mirrors (hf-mirror.com). It also provides instructions to install the library directly from the k2-fsa organization's GitHub repository. These are standard practices for AI model deployment from established research organizations.
[COMMAND_EXECUTION]: The skill documents the use of CLI tools like omnivoice-demo, omnivoice-infer, and omnivoice-infer-batch. These tools are part of the omnivoice package and are used for local inference and serving a web-based demonstration interface.
[SAFE]: Analysis of the Python API and code snippets shows no signs of credential harvesting, unauthorized file access, or hidden network exfiltration. The use of standard libraries like PyTorch and torchaudio is consistent with the skill's stated purpose of audio generation.

omnivoice-tts