This pull request significantly enhances the dataset card for `ExpressiveSpeech` by:
- Adding `language` (`en`, `zh`), `license` (`cc-by-nc-sa-4.0`), `task_categories` (`audio-classification`, `text-to-speech`), `tags` (e.g., `audio`, `speech`, `emotion`, `bilingual`, `tts`, `s2s`, `expressiveness`), and `size_categories` (`10K<n<100K`) to the YAML metadata for improved discoverability and searchability, while preserving existing dataset configuration information.
- Adding direct links to the associated paper ([huggingface.co/papers/2510.20513](https://huggingface.co/papers/2510.20513)) and GitHub repository ([github.com/FreedomIntelligence/ExpressiveSpeech](https://github.com/FreedomIntelligence/ExpressiveSpeech)) at the top of the dataset card.
- Integrating the paper abstract to provide immediate context about the dataset's purpose and methodology.
- Including a "Sample Usage" section with practical code snippets for inference, directly extracted from the project's GitHub README.
T…
