A TensorFlow Implementation of DC-TTS: yet another text-to-speech model

Implementation of Efficiently Trainable Text-to-Speech System Based on Deep Convolutional Networks with Guided Attention. Original code taken from here.

Requirements

• NumPy >= 1.11.1
• TensorFlow >= 1.3 (Note that the API of tf.contrib.layers.layer_norm has changed since 1.3)
• librosa
• tqdm
• matplotlib
• scipy

Try it

Check this colab to train the model with a dataset of yours.