A PyTorch implementation of Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech Synthesis

Add support for blizzard dataset.

pip3 install -r requirements.txt

• Hyperparameters.py --- hyperparameters
• Network.py --- encoder and decoder
• Modules.py --- some modules for tacotron
• Loss.py --- loss function
• Data.py --- dataset loader
• utils.py --- some util functions for data I/O
• Synthesis.py --- speech generation

• Download a multispeaker dataset
• Preprocess your data and implement your get_XX_data function in Data.py
• Set hyperparameters in Hyperparameters.py
• Make a directory named log as follow:

--- log
|    |
|    --- log[log_number]
|
--- code
     |
     --- Tacotron
             |
             --- train.py
             |
             --- Network.py
             |
           ......

• Run train.py

python3 train.py [log_number] [dataset_size] [start_epoch]

[log_number]: the log directory number
[dataset_size]: int or all
[start_epoch]: which epoch start to train (0 if start from scratch )

for example:
python3 train.py 0 all 0

Rungenerate.py. Replace the text in generate.py with any chinese sentences as you like before running

The pretained model provided is trained on Chinese dataset, so it only supports chinese now.