KuralHub lists 2 speech emotion recognition datasets for Kurdish. Each entry below gives the emotion labels, size, recording setup, licence, where to get it and how to cite it.
This database includes 7 emotions (Anger, Happiness, Sad, Fear, Boredom, Surprise, and Neutral) acted by 6 male and 6 female actors. Each actor utters 10 Kurdish sentences on 4 different sessions for each emotion. There is a total of 3360 recordings (i.e.,12×7×10×4 recordings). The speakers have been told to act these sentences and express them in their own style, by remembering/imagining a situation with the relevant emotion. The speakers have different acting experience (2 to 8 Years) and ages (19 to 36 years). The recording process has been done in a quiet room without restriction in the recording path. 10 listeners participated in a subjective test to determine the perceptibility of the emotions. The average of correct labeling was 41%.
Citation
@inproceedings{al2013excitation,
title={Excitation source and low level descriptor features fusion for emotion recognition using SVM and ANN},
author={Al-Talabani, Abdulbasit and Sellahewa, Harin and Jassim, Sabah},
booktitle={2013 5th computer science and electronic engineering conference (CEEC)},
pages={156--161},
year={2013},
organization={IEEE}
}
The Sorani Kurdish Speech Emotion Dataset was developed to facilitate research in emotion recognition within the Sorani dialect of the Kurdish language. This dataset addresses the scarcity of resources available for Kurdish, particularly in the context of speech emotion recognition.
Purpose: The primary goal of this dataset is to enable the development and evaluation of machine learning models capable of accurately identifying emotions in Sorani Kurdish speech. Such models have applications in various domains, including human-computer interaction, mental health monitoring, and call center analytics.
Collection Method: The dataset comprises speech samples meticulously collected from publicly accessible videos on the YouTube platform. These samples were carefully selected to represent four distinct emotions: neutral, sadness, happiness, and anger. Each audio segment was annotated according to these emotional categories, ensuring the dataset's relevance for training and testing emotion recognition systems.
By providing this specialized dataset, the study aims to bridge the gap in resources for Kurdish language processing and promote advancements in emotion recognition technologies for underrepresented languages.
Citation
@article{nematullah2024emotion,
title={Emotion Recognition in Kurdish Speech from the Sorani Dialect Corpus},
author={Nematullah, Omar and Askar, Shavan and Wahhab, Shahab and Khidir, Bzar},
journal={Zanco Journal of Pure and Applied Sciences},
volume={36},
number={5},
pages={104--112},
year={2024}
}
Know another Kurdish speech emotion dataset, or a newer version of one listed here? Suggest it or open a pull request.
This page is generated from datasets/Kurdish/ in the KuralHub repository.
If you use a dataset, cite its authors. If KuralHub helped you find it, please also cite the KuralHub paper.