Upload README.md with huggingface_hub

README.md

@@ -1,56 +1,82 @@
-# Uyghur TTS Dataset
+---
+license: cc0-1.0
+task_categories:
+  - text-to-speech
+language:
+  - ug
+tags:
+  - uyghur
+  - tts
+  - speech-synthesis
+  - common-voice
+size_categories:
+  - 10K<n<100K
+---
 
-A high-quality Uyghur text-to-speech dataset derived from Common Voice.
+# Uyghur Common Voice TTS Dataset
 
-## Dataset Overview
+A cleaned and processed Text-to-Speech (TTS) dataset for the Uyghur language, derived from Mozilla Common Voice.
 
-| Metric | Value |
-|--------|-------|
+## Dataset Summary
+
+| Property | Value |
+|----------|-------|
+| Language | Uyghur (ug) |
 | Total Samples | 43,054 |
-| Total Duration | 81.43 hours |
-| Average Duration | 6.81 seconds |
-| Train Set | 40,901 samples |
-| Validation Set | 2,153 samples |
+| Train Samples | 40,901 |
+| Validation Samples | 2,153 |
+| Audio Format | WAV |
+| Source | Mozilla Common Voice |
+| License | CC0-1.0 |
 
-## Directory Structure
+## Dataset Structure
 
 ```
-├── train.jsonl          # Training set
-├── val.jsonl            # Validation set
-├── wavs/                # Audio files (43,054 WAV files)
-└── ref_audio/           # Reference audio
+/
+├── train.jsonl          # Training data (40,901 samples)
+├── val.jsonl            # Validation data (2,153 samples)
+├── wavs/               # Audio files (43,054 WAV files)
+│   └── common_voice_ug_*.wav
+├── ref_audio/
+│   └── reference.wav   # Reference audio for voice cloning
+└── README.md
 ```
 
 ## Data Format
 
-Each line in the JSONL files contains:
-
-```json
-{
-  "audio": "wavs/common_voice_ug_xxx.wav",
-  "text": "سىلەردىكى مەسىلە دەل سىلەر ۋاز كەچكەن نەرسىلەرگە بولغان قىيماسلىق تۇيغۇسىدۇر.",
-  "g2p_text": "silɛrdiki mɛsilɛ dɛl silɛr waz kɛt͡ʃkɛn nɛrsilɛrɡɛ bolʁan qijmasliq tujʁusidur.",
-  "ref_audio": "ref_audio/reference.wav"
-}
-```
+Each line in `train.jsonl` / `val.jsonl` is a JSON object with:
 
 | Field | Description |
 |-------|-------------|
-| audio | Path to the audio file |
-| text | Uyghur text (Arabic script) |
-| g2p_text | Latin transliteration (G2P format) |
-| ref_audio | Reference audio path |
+| `audio` | Path to the WAV audio file |
+| `text` | Transcription text in Uyghur script |
+| `ref_audio` | Path to reference audio file |
+| `g2p_text` | Phonetic transcription (IPA format) |
 
-## Audio Specifications
+### Example
 
-- Format: WAV
-- Duration range: 2.16 - 15.48 seconds
-- Most audio (74.9%) is between 5-10 seconds
+```json
+{"audio": "wavs/common_voice_ug_39468550.wav", "text": "مېنىڭدە خېلى مال", "ref_audio": "ref_audio/reference.wav", "g2p_text": "meniŋdɛ χeli mal"}
+```
+
+## Usage
 
-## Source
+```python
+from datasets import load_dataset
 
-Data derived from [Mozilla Common Voice](https://commonvoice.mozilla.org/) Uyghur dataset.
+dataset = load_dataset("anke01/uyghur-common-voice-tts")
+print(f"Train: {len(dataset['train'])} samples")
+print(f"Validation: {len(dataset['validation'])} samples")
+```
 
-## License
+## Citation
 
-Same as Common Voice: CC0-1.0
+```bibtex
+@misc{uyghur-common-voice-tts,
+  author = {anke01},
+  title = {Uyghur Common Voice TTS Dataset},
+  year = {2026},
+  publisher = {Hugging Face},
+  url = {https://huggingface.co/datasets/anke01/uyghur-common-voice-tts}
+}
+```