File size: 626 Bytes
859a3bb 085185b bee89d5 085185b da6bd90 bee89d5 da6bd90 bee89d5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 |
---
license: cc-by-sa-4.0
tags:
- speaker extraction
- speech separation
- lip
- face
- multimodal
- cocktail party problem
---
A modified version of the VoxCeleb2 Dataset. Original data can be downloaded [here](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox2.html).
This dataset is used for Audio-visual speaker extraction conditioned on face recordings in [the reentry paper](https://ieeexplore.ieee.org/document/9721129), which the code can be found [here](https://github.com/zexupan/reentry).
# Usage
``` sh
cat orig* > orig.tar
tar -xvf orig.tar
cat audio_clean* > audio_clean.tar
tar -xvf audio_clean.tar
```
|