Update README.md

README.md

@@ -1,3 +1,43 @@
----
-license: cc-by-4.0
----
+---
+license: cc-by-4.0
+---
+
+# 🎬 ScreenTalk-XS: Sample Speech Dataset from Movies & TV 📺
+
+![Hugging Face](https://img.shields.io/badge/HuggingFace-Dataset-blue)  
+![License: CC BY-NC 4.0](https://img.shields.io/badge/License-CC%20BY--NC%204.0-orange)  
+![Dataset Type: XS (Limited Version)](https://img.shields.io/badge/Access-Limited-green)  
+
+### 📢 **What is ScreenTalk-XS?**
+**ScreenTalk-XS** is a **limited sample** from the full **ScreenTalk dataset**, containing **transcribed speech data** from selected movie and TV dialogues.  
+It is designed for **automatic speech recognition (ASR), natural language processing (NLP), and conversational AI research**.  
+
+✅ **This dataset serves as a free preview** of the full **ScreenTalk** dataset, which includes a much larger collection of movie and TV speech data.  
+
+---
+
+## 📜 **Dataset Details**
+- **Total Clips**: ~10% of the full dataset  
+- **Languages**: English  
+- **Content**: Transcribed movie & TV dialogues  
+- **Format**: `.wav` (audio) + `.tsv` (transcriptions)  
+- **Use Case**: Speech recognition, NLP, AI model training  
+
+🆓 **The full dataset (`ScreenTalk`) contains train, test, and validation splits and is available with restricted access.**
+
+---
+
+## 📂 **Dataset Structure**
+ScreenTalk-XS contains the following fields:
+
+| Column | Description |
+|--------|-------------|
+| `audio` | Audio file path |
+| `sentence` | Transcribed speech |
+
+📌 **Example Entry**
+```json
+{
+    "audio": "path/to/audio.wav",
+    "sentence": "I will find you and I will train my model."
+}