A Python-based modular toolbox for building Deep Neural Network models (using PyTorch) for statistical parametric speech synthesis - View it on GitHub