aranemini/central-kurdish-tts4all

TTS4All Central Kurdish Speech Dataset

Dataset Summary

The TTS4All Central Kurdish Speech Dataset is a multi-speaker speech corpus developed for speech synthesis and speech technology research in Central Kurdish (Sorani Kurdish).

The dataset was created within the TTS4All initiative during the JSALT 2025 Workshop and provides more than 35 hours of transcribed speech from three native Central Kurdish speakers.

The corpus was designed to support:

• Text-to-Speech (TTS)
• Voice Cloning
• Speech Translation
• Automatic Speech Recognition (ASR)
• Speech Data Augmentation
• Low-Resource Speech Technology Research

Dataset Statistics

Speaker	Gender	Domain	Utterances	Duration
fatih	Male	Audiobook	6,044	10h51m
shahen	Female	Audiobook	5,572	10h54m
giganet	Male	Studio Speech	6,055	13h35m
Total	—	—	17,671	35h20m

Dataset Structure

.
├── metadata.csv
└── wavs
    ├── fatih
    ├── shahen
    └── giganet

Metadata Format

file_name,text,speaker
wavs/fatih/sherko_01_9.wav,هاتینە کتێبی دووهەمی نووسین بە ئاوی خۆڵەمێش و بەشی بیستەم.,fatih
wavs/fatih/sherko_01_10.wav,وە بەڵام هەر وەک دەزانن نەزانی پیسترین نەخۆشییە، کە سێبەری ڕەشی خۆی خستۆتە سەر باڵای مێژووی مرۆڤایەتی.,fatih
wavs/fatih/sherko_01_11.wav,وەرن با هەوڵ بدەین بە خوێندنەوەی کتێب مێشک و بیر و بڕوامان پتەووتر و دەوڵەمەندتر بکەین.,fatih

Columns

Column	Description
audio	Relative path to the audio file
text	Central Kurdish transcription
speaker	Speaker identifier (fatih, shahen, or giganet)

Audio Information

• Format: WAV
• Sampling Rate: 24 kHz
• Channels: Mono
• Language: Central Kurdish (Sorani Kurdish)
• Script: Arabic-based Kurdish script

Dataset Creation

The dataset contains speech from three speakers:

fatih

Male audiobook narrator.

• 6,044 utterances
• 10h51m

shahen

Female audiobook narrator.

• 5,572 utterances
• 10h54m

giganet

Male studio-recorded speech.

• 6,055 utterances
• 13h35m

The audiobook datasets were created by:

1. Segmenting long audiobook recordings using Voice Activity Detection (VAD).
2. Automatically transcribing the speech using a fine-tuned Central Kurdish ASR system.
3. Manually revising and validating transcriptions.

Intended Uses

This dataset is intended for:

• Text-to-Speech training
• Voice cloning
• Speech translation
• Automatic speech recognition
• Speech data augmentation
• Low-resource speech technology research

Associated Models

This dataset was used to train three Central Kurdish F5-TTS models:

Model	Speaker
audiobook-F	shahen
audiobook-M	fatih
studio-M	giganet

Language Information

Central Kurdish (Sorani Kurdish)

• ISO 639-3: ckb
• Script: Arabic-based Kurdish alphabet

License

This dataset is released under the CC BY-NC-ND 4.0 license.

Restrictions

• Commercial use is prohibited.
• Redistribution of modified versions is prohibited.
• The dataset may only be used for research, educational, and non-commercial purposes.

Acknowledgements

This work was developed within the TTS4All project at the JSALT 2025 Workshop and at LIUM, Le Mans University.

We thank:

• Malaka Mustafa Soltani for granting permission to use the audiobook content.
• Shahen Shahlai for allowing the use of her voice recordings.
• Fatih for allowing the use of his voice recordings.
• The volunteers who participated in the evaluation campaigns.

Citation

If you use this dataset, please cite:

@inproceedings{mohammadamini-etal-2026-central,
  title = {Central Kurdish Text-to-Speech and Its Application in Speech-to-Text Translation},
  author = {Mohammadamini, Mohammad and Shamsi, Meysam and Tahon, Marie},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026)},
  month = {May},
  year = {2026},
  pages = {664--673},
  address = {Palma, Mallorca, Spain},
  publisher = {European Language Resources Association (ELRA)},
  editor = {Piperidis, Stelios and Bel, Núria and van den Heuvel, Henk and Ide, Nancy and Krek, Simon and Toral, Antonio},
  doi = {10.63317/4hfwowidu34u}
}

@inproceedings{mohammadamini2026iwslt,
  title        = {LIUM Submission for IWSLT 2026 Low-Resource Speech Translation Track},
  author       = {Mohammad Mohammadamini and Marie Tahon},
  year         = {2026},
  howpublished = {Proceedings of the International Conference on Spoken Language Translation (IWSLT) 2026},
}