boboyiyi/multi_speaker_tacotron

A TensorFlow implementation of multi speaker Tacotron speech synthesis

multi_speaker_tacotron

Quick Start

Installing dependencies

1. Running environment: tensorflow 1.4.0 + python 3.6

2. pip install -r requirements.txt

Training

Preprocess the data

python preprocess.py --dataset [your dataset]

Train a model

Before running, you need to change the default configuration of hparams.py according to your dataset.

sh run_train.sh

Synthesize from a checkpoint

sh run_eval.sh

References:

• https://github.com/keithito/tacotron
• https://github.com/kastnerkyle/multi-speaker-tacotron-tensorflow

To do:

• Upload source code.
• Run and verify.
• Detailed readme.