metadata

license: cc-by-4.0
size_categories:
  - 1K<n<10K
task_categories:
  - audio-classification
tags:
  - speaker
  - height

HeightCeleb

This repository contains the dataset described in article "HeightCeleb - an enrichment of VoxCeleb dataset with speaker height information", which was presented at SLT 2024 conference in Macau, China.

The dataset is an extension to Voxceleb dataset and contains height information that was scraped from the Internet.

Live demo of system trained with HeightCeleb

Demo system deployed in HuggingFace Spaces

Dataset Attribution

HeightCeleb dataset is an extension of Voxceleb dataset.

A. Nagrani, J. S. Chung, A. Zisserman, "VoxCeleb: a large-scale speaker identification dataset", INTERSPEECH, 2017

Original dataset: https://www.robots.ox.ac.uk/~vgg/data/voxceleb/
License: https://www.robots.ox.ac.uk/~vgg/data/voxmovies/files/license.txt

Modifications

This dataset contains VoxCeleb1 ID, sex and split information from the Voxceleb dataset.

Licenese

HighCeleb dataset is distributed under CC BY 4.0 license.

Citing HeightCeleb

If you use HeightCeleb in your research, please cite it using the following BibTeX entry:

@INPROCEEDINGS{10832224,
  author={Kacprzak, Stanisław and Kowalczyk, Konrad},
  booktitle={2024 IEEE Spoken Language Technology Workshop (SLT)}, 
  title={Heightceleb - An Enrichment of Voxceleb Dataset With Speaker Height Information}, 
  year={2024},
  volume={},
  number={},
  pages={857-862},
  doi={10.1109/SLT61566.2024.10832224}}