File size: 955 Bytes
a537441 b64c23d b0eb7e1 b64c23d 1dbb1a3 b64c23d |
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 26 27 28 29 30 |
---
license: cc-by-nc-nd-4.0
task_categories:
- automatic-speech-recognition
language:
- fr
size_categories:
- 100K<n<1M
---
The french subset of the dataset [Multilingual TEDx](https://www.openslr.org/100). The data uploaded to HF corresponds to the directory fr-fr
Configs:
- single_samples (default): all samples taken separately
- max=30s: combine consecutive samples for a period shorter than 30 seconds
- max=10s: combine consecutive samples for a period shorter than 10 seconds
- max: combine all the samples of a TEDx talk
```
dependencies:
- ffmpeg: apt install ffmpeg
- ffmpeg-python: pip install ffmpeg-python
@inproceedings{salesky2021mtedx,
title={Multilingual TEDx Corpus for Speech Recognition and Translation},
author={Elizabeth Salesky and Matthew Wiesner and Jacob Bremerman and Roldano Cattoni and Matteo Negri and Marco Turchi and Douglas W. Oard and Matt Post},
booktitle={Proceedings of Interspeech},
year={2021},
}
``` |