zldzmfoq12 / Prosody-Tacotron

Prosody-Tacotron

Korean Speech Synthesis with Prosody-Tacotron which based on both of Tacotron1 and Tacotron2 model

Note that this repo is based on https://github.com/syang1993/gst-tacotron

Background

In March 2018, Google published a paper, [Towards End-to-End Prosody Transfer for Expressive Speech Synthesis with Tacotron], where they present a neural text-to-speech model that learns to synthesize speech directly from (text, audio) pairs. However, they didn't release their source code or training data. This is an independent attempt to provide an open-source implementation of the model described in their paper.

The quality isn't as good as Google's demo yet, but hopefully it will get there someday :-). Pull requests are welcome!

Quick Start

Installing dependencies

1. Install Python 3.

2. Install the latest version of TensorFlow for your platform. For better performance, install with GPU support if it's available. This code works with TensorFlow 1.3 and later.

3. Install requirements:

   pip install -r requirements.txt
   

Training

Note: you need at least 40GB of free disk space to train a model.

1. Download a speech dataset.

   The following are supported out of the box:

   • KSS Dataset (Public Domain)

   You can use other datasets if you convert them to the right format. See TRAINING_DATA.md for more info.

2. Unpack the dataset into ~/tacotron

   After unpacking, your tree should look like this for LJ Speech:

   tacotron
     |- kss
         |- metadata.csv
         |- wavs
   

3. Preprocess the data

   python3 preprocess.py --dataset kss
   
   

4. Train a model

   python3 train.py
   

   • you can choose tacotron model between Tacotron1 and Tacotron2 by model in hparams.py
   • Tunable hyperparameters are found in hparams.py. You can adjust these at the command line using the --hparams flag, for example --hparams="model=tacotron2,gst_index=2". Hyperparameters should generally be set to the same values at both training and eval time. The default hyperparameters are recommended for LJ Speech and other English-language data. See TRAINING_DATA.md for other languages.

5. Monitor with Tensorboard (optional)

   tensorboard --logdir ~/tacotron/logs-tacotron
   

   The trainer dumps audio and alignments every 1000 steps. You can find these in ~/tacotron/logs-tacotron-attention_type.

6. Synthesize from a checkpoint

   python3 demo_server.py --checkpoint ~/tacotron/logs-tacotron/model.ckpt-185000
   

   Replace "185000" with the checkpoint number that you want to use, then open a browser to localhost:9000 and type what you want to speak. Alternately, you can run eval.py at the command line:

   python3 eval.py --checkpoint ~/tacotron/logs-tacotron/model.ckpt-185000 --reference_audio /path/to/ref_audio
   

   • If you don't use the --reference_audio, you can select and scale style by gst_index, gst_scale in hparams.py
   • If you set the --hparams flag when training, set the same value here.