galsenai/wolof_tts

---
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: gender
dtype: string
splits:
- name: train
num_bytes: 4274917077.83
num_examples: 36009
- name: test
num_bytes: 475541653.87
num_examples: 4001
download_size: 4571561616
dataset_size: 4750458731.7
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
license: cc-by-4.0
task_categories:
- text-to-speech
language:
- wo
pretty_name: woloftts
size_categories:
- 10K<n<100K
---
# Wolof TTS
## Dataset Description
This is a [Wolof](https://en.wikipedia.org/wiki/Wolof_language) Text To Speech (TTS) dataset collected by [Baamtu Datamation](https://baamtu.com/) as part of the [AI4D African language program](https://k4all.org/project/language-dataset-fellowship/).
The original dataset is hosted on [Zenodo](https://zenodo.org/records/4498861#.YXU2A3X7R-M) and it contains recordings from two (02) natif Wolof speakers (a `male` and `female` voice). Each speaker recored __more than 20,000 sentences__.

### Speaking time:

-- Male: 22h 28mn 41s
-- Female: 18h 47mn 19s

The text dataset comes from _news websites_, _Wikipedia_ and _self curated text_.

## Citation
You can access the project paper at https://arxiv.org/abs/2104.02516.
If you work on the dataset, please cite the authors below.
```
@dataset{thierno_ibrahima_diop_2021_4498861,
author = {Thierno Ibrahima Diop and
Demba AW and
Ami jaane and
Mamadou Badiane},
title = {WOLOF TTS(Text To Speech) Data},
month = feb,
year = 2021,
publisher = {Zenodo},
version = 1,
doi = {10.5281/zenodo.4498861},
url = {https://doi.org/10.5281/zenodo.4498861}
}
```
This dataset was made available on HuggingFace through [GalsenAI](https://galsen.ai/)'s mission to strengthen Senegal's AI ecosystem. It was cleaned and structured by [Derguene Mbaye](https://huggingface.co/derguene), then uploaded and documented by [Alwaly Ndiaye](https://huggingface.co/Alwaly).