lightsunny/multivsr
收藏Hugging Face2026-03-07 更新2026-03-29 收录
下载链接：
https://hf-mirror.com/datasets/lightsunny/multivsr
下载链接
链接失效反馈官方服务：
资源简介：
---
configs:
- config_name: test
data_files: test.csv
- config_name: train_val_test
data_files: '*.csv'
- config_name: metadata
data_files: multivsr.tar
license: mit
language:
- en
- fr
- de
- es
- it
- ca
- ru
- ja
- zh
- pl
- pt
- tr
- nl
tags:
- lipreading
- audiovisual
- video
- asr
- avsr
- talkingface
- audio
- speech
---
# Dataset: MultiVSR
We introduce a large-scale multilingual lip-reading dataset: MultiVSR. The dataset comprises a total of 12,000 hours of video footage, covering English + 12 non-English languages. MultiVSR is a massive dataset with a huge diversity in terms of the speakers as well as languages, with approximately 1.6M video clips across 123K YouTube videos. Please check the [website](https://www.robots.ox.ac.uk/~vgg/research/multivsr/) for samples.
<p align="center">
<img src="dataset_teaser.gif" alt="MultiVSR Dataset Teaser">
</p>
## Download instructions
Please check the GitHub repo to download, preprocess, and prepare the dataset: https://github.com/Sindhu-Hegde/multivsr/tree/master/dataset.
提供机构：
lightsunny
