readme.md

Fongbe Speech Collection

2th place of Google NLP Hack Series

Data

You must first download the dataset by following this link and unzip the file by putting the lab and wav subfolders at the root of the dataset folder.

Train a model

Requirements

pip install -r requirements.txt

Run training

To train a model with our collected data, run the following

python train.py

Loss and WER obtained

Citation

Cite as:

@dataset{laleye_frejus_a_a_2022_6604637,
  author       = {Laleye Fréjus A. A.},
  title        = {Fongbe speech dataset},
  month        = {jun},
  year         = {2022},
  publisher    = {Zenodo},
  version      = {1.0},
  doi          = {10.5281/zenodo.6604637},
  url          = {https://doi.org/10.5281/zenodo.6604637}
}