KuralHub lists 2 speech emotion recognition datasets for French, 2 of them openly available. Each entry below gives the emotion labels, size, recording setup, licence, where to get it and how to cite it. French is also part of the benchmark. The best of 11 models reached 70.0% test accuracy (wavlm-large).
French 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 42.1% (median 38.0%). The best run reached 70.0% with wavlm-large. Compare with other languages.
CaFE (run ID CaFE). Scores on the test split, best run highlighted.
The Canadian French Emotional (CaFE) speech dataset focuses on emotional speech in Canadian French, featuring six sentences spoken by 12 actors (6 male, 6 female) across six basic emotions plus neutral speech. Each emotion is expressed in two intensity levels (mild and strong). The dataset is available in high-resolution (192 kHz, 24-bit) and a downsampled version (48 kHz, 16-bit) under a Creative Commons (CC BY-NC-SA 4.0) license.
Used in the KuralHub benchmark (run ID CaFE). See the results above.
Citation
@inproceedings{gournay2018canadian,
title={A canadian french emotional speech dataset},
author={Gournay, Philippe and Lahaie, Olivier and Lefebvre, Roch},
booktitle={Proceedings of the 9th ACM multimedia systems conference},
pages={399--402},
year={2018}
}
The Oréau French emotional speech database was recorded in a quiet environment to support research on emotional speech and speech synthesis. It includes 79 classroom-relevant utterances covering 7 emotions, spoken by 32 non-professional speakers. Two versions exist: the first with 502 utterances, and the second with 434 utterances retained after a perception test confirmed 90% accuracy in emotion recognition and naturalness.
Citation
@misc{kerkeni2020french,
title={French emotional speech database-or{\'e}au},
author={KERKENI, Leila and CLEDER, Catherine and Youssef, SER-RESTOU and RAOOF, Kosai},
year={2020},
publisher={Dec}
}
Know another French speech emotion dataset, or a newer version of one listed here? Suggest it or open a pull request.
This page is generated from datasets/French/ in the KuralHub repository.
If you use a dataset, cite its authors. If KuralHub helped you find it, please also cite the KuralHub paper.