KuralHub lists 2 speech emotion recognition datasets for Indonesian, 1 of them openly available. Each entry below gives the emotion labels, size, recording setup, licence, where to get it and how to cite it. Indonesian is also part of the benchmark. The best of 11 models reached 99.0% test accuracy (hubert-base-ls960).
Indonesian is one of the 29 languages in the KuralHub benchmark. Each of the 11 frozen speech models was trained with a small classification head on the shared emotions and scored on a held-out test split. Over all 11 runs the mean test accuracy was 72.1% (median 77.0%). The best run reached 99.0% with hubert-base-ls960. Compare with other languages.
IndoWaveSentiment (run ID IndoWaveSentiment). Scores on the test split, best run highlighted.
The dataset consists of audio recordings extracted from the Indonesian widescreen movie Cek Toko Sebelah. It was chosen for its well-structured dialogues, high audio quality, and minimal background noise. The dataset is designed to limit variations in speakers, accents, and recording techniques by using a single movie, ensuring consistency. This approach is similar to past studies that controlled variations by using a limited number of speakers and sentences per emotion class. However, restricting the dataset to one film also limits the total amount of data collected.
Citation
@inproceedings{fahmi2020speech,
title={Speech-Emotion Detection in an Indonesian Movie},
author={Fahmi, Fahmi and Jiwanggi, Meganingrum Arista and Adriani, Mirna},
booktitle={Proceedings of the 1st Joint Workshop on Spoken Language Technologies for Under-resourced languages (SLTU) and Collaboration and Computing for Under-Resourced Languages (CCURL)},
pages={185--193},
year={2020}
}
IndoWaveSentiment is an Indonesian speech dataset for emotion classification, featuring 300 mono-channel 16 kHz audio recordings from 10 actors (5 male, 5 female). Each actor repeated a sentence three times across five emotions: neutral, happy, surprised, disgusted, and disappointed. Designed for deep learning and signal processing, it supports research in sentiment analysis and voice emotion classification.
Used in the KuralHub benchmark (run ID IndoWaveSentiment). See the results above.
Citation
@article{bustamin2024indowavesentiment,
title={IndoWaveSentiment: Indonesian audio dataset for emotion classification},
author={Bustamin, Anugrayani and Rizky, Andi M and Warni, Elly and Areni, Intan Sari and others},
journal={Data in Brief},
volume={57},
pages={111138},
year={2024},
publisher={Elsevier}
}
Know another Indonesian speech emotion dataset, or a newer version of one listed here? Suggest it or open a pull request.
This page is generated from datasets/Indonesian/ in the KuralHub repository.
If you use a dataset, cite its authors. If KuralHub helped you find it, please also cite the KuralHub paper.