I was trying to do some synthesis with trained models.
By doing “CUDA_VISIBLE_DEVICES=0 python synthesize.py “To be or not to be” config.json …/keep/best_model.pth.tar …/keep/ --use_cuda true”
which follows the input of text, config, model_path, out_path, use_cuda
I got this error: AttributeError: ‘NoneType’ object has no attribute ‘_normalize’
The model was trained by Tacotron2 from master branch.
Anything I could do about this? It looks like something wrong with source code.