Update README.md

README.md

@@ -6,4 +6,20 @@ 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
+
+Splits:
+- 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},
+}
+```