Update README.md
Browse files
README.md
CHANGED
@@ -6,4 +6,20 @@ language:
|
|
6 |
- fr
|
7 |
size_categories:
|
8 |
- 100K<n<1M
|
9 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
- fr
|
7 |
size_categories:
|
8 |
- 100K<n<1M
|
9 |
+
---
|
10 |
+
The french subset of the dataset [Multilingual TEDx](https://www.openslr.org/100). The data uploaded to HF corresponds to the directory fr-fr
|
11 |
+
|
12 |
+
Splits:
|
13 |
+
- single_samples (default): all samples taken separately
|
14 |
+
- max=30s: combine consecutive samples for a period shorter than 30 seconds
|
15 |
+
- max=10s: combine consecutive samples for a period shorter than 10 seconds
|
16 |
+
- max: combine all the samples of a TEDx talk
|
17 |
+
|
18 |
+
```
|
19 |
+
@inproceedings{salesky2021mtedx,
|
20 |
+
title={Multilingual TEDx Corpus for Speech Recognition and Translation},
|
21 |
+
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},
|
22 |
+
booktitle={Proceedings of Interspeech},
|
23 |
+
year={2021},
|
24 |
+
}
|
25 |
+
```
|