This PR significantly improves the model card for `MeanVC` by:
- **Correcting Metadata**: The `pipeline_tag` is updated from the incorrect `text-to-speech` to `audio-to-audio` to accurately reflect its functionality as a voice conversion model. The erroneous `text-to-speech` tag is also removed.
- **Adding Comprehensive Sample Usage**: A new "Sample Usage" section is added, directly incorporating the detailed "Getting Started" instructions (environment setup, model download, real-time and offline conversion code snippets) from the project's GitHub README. This makes it much easier for users to interact with and use the model.
- **Updating Image Paths**: Relative image paths (`figs/model.png`, `figs/npu@aslp.jpeg`) are updated to absolute URLs on the Hugging Face Hub (`https://huggingface.co/ASLP-lab/MeanVC/resolve/main/figs/model.png`) for improved robustness and rendering.
These changes enhance the model's discoverability and provide users with clearer, more actionable information.
