Voice conversion software - Voice conversion (VC) is a technique to convert a speaker identity of a source speaker into that of a target speaker. This software enables the users to develop a traditional VC system based on a Gaussian mixture model (GMM) and a vocoder-free VC system based on a differential GMM (DIFFGMM) using a parallel dataset of the source and target speakers.
- Voice Conversion Challenge 2018 [zip]
This software was developed to make it possible for the users to easily build the VC systems by only preparing a parallel dataset of the desired source and target speakers and executing example scripts. The following VC methods were implemented as the typical VC methods.
- T. Toda, A.W. Black, K. Tokuda, "Voice conversion based on maximum likelihood estimation of spectral parameter trajectory," IEEE Transactions on Audio, Speech and Language Processing, Vol. 15, No. 8, pp. 2222-2235, Nov. 2007.
- K. Kobayashi, T. Toda, S. Nakamura, "F0 transformation techniques for statistical voice conversion with direct waveform modification with spectral differential," Proc. IEEE SLT, pp. 693-700, Dec. 2016.
To make it possible to easily develop VC-based applications using Python (Python3), the VC library is also supplied, including several interfaces, such as acoustic feature analysis/synthesis, acoustic feature modeling, acoustic feature conversion, and waveform modification. For the details of the VC library, please see sprocket documents in (coming soon).
Please use NOT Python2 BUT Python3.
Ver. 0.18
pip install numpy # for dependency
pip install -r requirements.txt
python setup.py install
See VC example
For any questions or issues please visit:
https://github.com/k2kobayashi/sprocket/issues
Copyright (c) 2017 Kazuhiro KOBAYASHI
Released under the MIT license
https://opensource.org/licenses/mit-license.php
Thank you @r9y9 and @tats-u for lots of contributions and encouragement helps before release.
-
Kazuhiro Kobayashi @k2kobayashi [maintainer, design and development]
-
Tomoki Toda [advisor]