|
--- |
|
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 |
|
|
|
``` |
|
@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}, |
|
} |
|
``` |
