---
dataset_info:
  features:
  - name: audio
    dtype: audio
  - name: duration
    dtype: float64
  - name: text
    dtype: string
  - name: tr
    dtype: string
  - name: N2gkPlus
    dtype: string
  - name: speaker
    dtype: string
  - name: LNCat
    dtype: string
  - name: en_convertable
    dtype: bool
  - name: Jamo Bigram
    dtype: string
  splits:
  - name: train
    num_bytes: 234581859846.24
    num_examples: 168790
  download_size: 189523270766
  dataset_size: 234581859846.24
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
pretty_name: CoreaSpeech
size_categories:
- n<1K
---


## 📘 Dataset Description

This dataset is part of a Korean multispeaker speech corpus project for TTS research.  
It contains preprocessed speech–text pairs, metadata, and linguistic annotations for model training.

---

### 📂 External Datasets

The following publicly available datasets can be downloaded from the links below:

- **KSS (Korean Single Speaker Speech Dataset)**  
  🔗 [Bingsu/KSS_Dataset](https://huggingface.co/datasets/Bingsu/KSS_Dataset)

- **EMILIA (Elderly Emotional Speech Dataset)**  
  🔗 [amphion/Emilia-Dataset](https://huggingface.co/datasets/amphion/Emilia-Dataset)

Preprocessing scripts are available at:  
🔗 [https://github.com/CoreaSpeech/sourcecode](https://github.com/CoreaSpeech/sourcecode)

---

### ⚙️ Usage Policy

1. For research purposes only  
2. No secondary modification or redistribution  
3. Voice identity reconstruction prohibited  
4. Commercial use not allowed

---

### ⚖️ License & Attribution

This dataset is released strictly for research and academic purposes.