Rayhane-mamah/Tacotron-2
A TensorFlow implementation of DeepMind's Tacotron-2 neural network for text-to-speech synthesis.
Not currently ranked — collecting fresh signals.
star history
This repository implements the Tacotron-2 architecture combining a seq2seq encoder-decoder that predicts mel spectrograms from text with a Wavenet vocoder that converts those spectrograms into audio waveforms. It provides hyperparameters to reproduce the original paper results along with additional improvements that improve output quality in most cases. The project includes pre-trained checkpoints, training logs, and utilities for processing standard TTS datasets.
Frequently asked
- What is Rayhane-mamah/Tacotron-2?
- A TensorFlow implementation of DeepMind's Tacotron-2 neural network for text-to-speech synthesis.
- Is Tacotron-2 open source?
- Yes — Rayhane-mamah/Tacotron-2 is open source, released under the MIT license.
- What language is Tacotron-2 written in?
- Rayhane-mamah/Tacotron-2 is primarily written in Python.
- How popular is Tacotron-2?
- Rayhane-mamah/Tacotron-2 has 2.3k stars on GitHub.
- Where can I find Tacotron-2?
- Rayhane-mamah/Tacotron-2 is on GitHub at https://github.com/Rayhane-mamah/Tacotron-2.