Update README.md

README.md

@@ -1,3 +1,36 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+task_categories:
+- audio-classification
+language:
+- pt
+size_categories:
+- 10K<n<100K
+---
+
+# Dataset card
+
+<!-- Provide a quick summary of the dataset. -->
+
+This dataset includes ~80k samples of speech audio in Brazilian Portuguese. Samples have variable length ranging from 1 to 4 seconds, with a sampling rate of 16kHz. The metadata file includes speaker tags and corresponding labels for each sample, making it appropriate for speaker identification and speaker verification tasks.
+
+## Dataset Description
+
+Audio samples are taken from three bigger corpora: C-ORAL Brasil, NURC Recife and NURC SP. Please take into consideration the licenses details of the original corpora.
+All the recordings come from monologues, but there are brief moments of interruption by the researchers that conducted the recordings.
+The recordings comprise accents from three different states of Brazil: Belo Horizonte, São Paulo and Recife.
+The dataset is structured in a way that makes it possible to load with HF Audiofolder. In my experience, I noticed that it is easier to clone the dataset repo to your machine and then load it specifying the local data_dir than to load it directly from the hub using the dataset tag.
+
+## Dataset Sources
+
+<!-- Provide the basic links for the dataset. -->
+
+- **C-ORAL BRASIL:** [https://www.c-oral-brasil.org/c-oral-brasil-i.php]
+- **NURC Recife:** [https://fale.ufal.br/projeto/nurcdigital/index.php?action=home]
+- **NURC SP:** [http://tarsila.icmc.usp.br:8080/nurc/home]
+
+## Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
+
+The models trained with this dataset should not be used to intentionally create hostile or alienating environments for people.