Add metadata and improve model card

README.md

@@ -1,9 +1,17 @@
-# ESRT
+---
+library_name: transformers
+pipeline_tag: automatic-speech-recognition
+---
 
-## Bandwidth-Efficient and Privacy-Preserving Edge-Cloud Many-to-Many Speech Translation.
+# ESRT: Edge-cloud Speech Recognition and Translation
+
+This repository contains the weights for ESRT-4B, as presented in the paper [Bandwidth-Efficient and Privacy-Preserving Edge-Cloud Many-to-Many Speech Translation](https://huggingface.co/papers/2605.28642).
 
 ESRT supports many-to-many speech-to-text translation across **45 languages** (45 × 44 directions). It uses an edge-cloud split inference architecture to protect voice privacy and reduce bandwidth by transmitting only compressed acoustic features instead of raw audio.
 
+- **Paper:** [arXiv:2605.28642](https://arxiv.org/abs/2605.28642)
+- **Code:** [https://github.com/yxduir/esrt](https://github.com/yxduir/esrt)
+
 [![arXiv](https://img.shields.io/badge/arXiv-2605.28642-b31b1b.svg)](https://arxiv.org/abs/2605.28642)
 [![Hugging Face Models](https://img.shields.io/badge/%F0%9F%A4%97-Models-yellow "https://huggingface.co/yxdu")](https://huggingface.co/yxdu/ESRT-4B)
 
@@ -65,4 +73,4 @@ Training code will be open-sourced in a future release. Validated on:
       primaryClass={cs.AI},
       url={https://arxiv.org/abs/2605.28642}, 
 }
-```
+```