Hi! I'm Niels from the Hugging Face community science team.
This PR improves the dataset card by adding relevant metadata to the YAML section at the top, including the task category (`automatic-speech-recognition`), language tags (`en`, `zh`), and descriptive tags like `multimodal` and `avsr`. This helps users find your dataset more easily when filtering on the Hub.
The existing documentation is excellent, so I've kept all the original markdown content intact.
