Speech2Phone

This is the official implementation of the paper Speech2Phone: A Multilingual and Text Independent Speaker Identification Model

Speech2Phone is a multilingual, text-independent speaker identification system. In addition, the embeddings extracted from this model can be used to represent speakers in speech synthesis systems, speech cloning and voice transfer between languages.

In this repository the Paper directory has the implementation of all the experiments and topologies explored in the article.   The Speech2Phone directory presents the implementation and checkpoints of the best model of the article.

Colab Notebook Demos:

     Identification of speakers in Spanish

     Identification of speakers in Chinese spoken in Taiwan

Citation

@article{casanova2020speech2phone,
  title={Speech2Phone: A Multilingual and Text Independent Speaker Identification Model},
  author={Casanova, Edresson and Junior, Arnaldo Candido and Shulby, Christopher and da Silva, Hamilton Pereira and Cordeiro, Alessandro Ferreira and Guedes, Victor de Oliveira and Aluisio, Sandra Maria and others},
  journal={arXiv preprint arXiv:2002.11213},
  year={2020}
}