Modeling early phonetic acquisition from child-centered audio data

This is the git repository associated to our publication: Modeling early phonetic acquisition from child-centered audio data In this repository, you'll find all the necessary code for training a contrastive predictive coding (CPC) model from raw speech. Adapted and modified from the publication Unsupervised Pretraining Transfers well Across Languages, whose companion git repository can be found here.

How to use ?

1. Installation
2. Data preparation
3. Training and Evaluation

References

Main paper:

@article{lavechin2024modeling,
    title={Statistical learning models of early phonetic acquisition struggle with child-centered audio data},
  author={Lavechin, Marvin and De Seyssel, Maureen and M{\'e}tais, Marianne and Metze, Florian and Mohamed, Abdelrahman and Bredin, Herv{\'e} and Dupoux, Emmanuel and Cristia, Alejandrina},
  year={2024},
  publisher={Cognition}
}