DALiH-ANR/artsakh_hy

Artsakh Dataset (Audio–Transcription Alignment)

An audio–transcription alignment dataset for the Artsakh varieties (Stepanakert, Getashen, Hadrut) of Armenian.

Overview

  This dataset contains Armenian dialect speech recordings (Artsakh varieties: Stepanakert, Getashen, Hadrut) paired with aligned transcriptions. It is split into three subsets:  

• Train: 5,122 examples 
• Validation (dev): 140 examples 
• Test: 175 examples

Content

  Each example includes:  

• audio: a WAV audio file 
• transcription: the Armenian transcription text 
• duration: audio duration in seconds

 

Intended Use

  This dataset is intended for research and development in areas such as:  

• Automatic Speech Recognition (ASR) for Armenian dialects
• Forced alignment / audio–text alignment
• Phonetic and linguistic analysis of Artsakh Armenian (Stepanakert, Getashen, Hadrut)

 

Citation

  If you use this dataset in academic work, please cite it as follows:   [TBD]  

License

  This dataset is licensed under the Creative Commons Attribution–NonCommercial 4.0 International License (CC BY-NC 4.0). Commercial use is not permitted without prior written permission from the authors.

Acknowledgments

The DALiH project is funded by French National Research Agency ANR-21-CE38-0006.