Datasets:
Tasks:
Audio Classification
Modalities:
Audio
Formats:
soundfolder
Languages:
Urdu
Size:
< 1K
ArXiv:
License:
metadata
license: mit
language:
- ur
size_categories:
- n<1K
task_categories:
- audio-classification
URDU-Dataset
1. General information
URDU dataset contains emotional utterances of Urdu speech gathered from Urdu talk shows. It contains 300 utterances of four basic emotions: Angry, Happy, and Neutral. There are 38 speakers (27 male and 11 female). This data is created from YouTube. Speakers are selected randomly.
For more details about dataset, please refer the complete paper "Cross Lingual Speech Emotion Recognition: Urdu vs. Western Languages". https://arxiv.org/pdf/1812.10411.pdf