🌍 CS-FLEURS: A Massively Multilingual and Code-Switched Speech Dataset

📖 Overview

CS-FLEURS is a new dataset for developing and evaluating code-switched speech recognition and translation systems beyond high-resourced languages.

• 113 unique code-switched language pairs across 52 languages
• 300 hours of speech data, both read and synthetic

📊 Dataset Statistics

CS-FLEURS consists of the following subsets:

• Read-Test: 14 X-English pairs, read speech
• XTTS-Train: 16 X-English pairs, generative TTS
• XTTS-Test1: 16 X-English pairs, generative TTS
• XTTS-Test2: 60 {Arabic, Chinese, Hindi, Spanish}-X pairs, generative TTS
• MMS-Test: 45 X-English pairs, concatenative TTS

Statistic	Read-Test	XTTS-Train	XTTS-Test1	XTTS-Test2	MMS-Test
Duration (hours)	17	128	36	42	56
Tokens (words)	128k	889k	257k	300k	315k
Matrix Langs	14	16	16	4	45
Embedded Langs	1	1	1	15	1
Total CS Pairs	14	16	16	60	45
Same-Script Pairs	7	10	10	10	22
Distinct-Script Pairs	7	6	6	51	23

🔽 How to Download

Clone with Git LFS:

git lfs install
git clone https://huggingface.co/datasets/byan/cs-fleurs

Or load directly in 🤗 Datasets:

from datasets import load_dataset

data = load_dataset("byan/cs-fleurs")

📜 License

This dataset is licensed under CC BY 4.0 for Non-Commercial use.

🙏 Citation

If you use this dataset, please cite:

@inproceedings{yan2025cs,
  title={CS-FLEURS: A Massively Multilingual and Code-Switched Speech Dataset},
  author={Yan, Brian and Hamed, Injy and Shimizu, Shuichiro and Lodagala, Vasista and Chen, William and Iakovenko, Olga and Talafha, Bashar and Hussein, Amir and Polok, Alexander and Chang, Kalvin and others}
}