Implementation of the paper titled "Neural Voice Cloning with Few Samples" by Baidu link
A checkpoint for the encoder trained on 56k epochs with a loss of 0.0810 can be found in the checkpoints directory
The repository is only partially complete. https://github.com/SforAiDl/Neural-Voice-Cloning-With-Few-Samples describes how to train and generate voice samples using Speaker Adaptation approach. Link to generated voice samples can be found there as well. All the future commits will happen on https://github.com/SforAiDl/Neural-Voice-Cloning-With-Few-Samples repository.
The implementation of deep voice 3 was done from the following repository: https://github.com/r9y9/deepvoice3_pytorch
If you find the code in the repository useful, please cite it using:
@misc{chitlangia2021voicecloning,
author = {Chitlangia, Sharad and Rastogi, Mehul and Ganguly, Rijul},
title = {An Open Source Implementation of Neural Voice Cloning With Few Samples},
year = {2021},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {Available at \url{https://github.com/SforAiDl/Neural-Voice-Cloning-With-Few-Samples/} or \url{https://github.com/Sharad24/Neural-Voice-Cloning-With-Few-Samples/}},
}