TensorflowTTS, an open source based on Tensorflow 2 that supports several latest TTS models such as Tacotron2, MelGan, FastSpeech, etc., has finally begun supporting Microsoft FastSpeech2. FastSpeech2 shows similar performance to Transformer series TTS, but it reduces the training time by more than 2 times and inference is more than 150 times faster. Below is the TensorflowTTS github link.
It is an open source project made with Tensorflow 2 and has a nice subtitle “Real-Time State-of-the-art Speech Synthesis for Tensorflow 2”. The update is also very active, and TFlite support was added on 7/5, Multi-GPU training on 7/17, Korean TTS on 8/5, and Chinese TTS on 8/14.
In addition to the Korean implementation listed above, there is also a version made by Professor Injoong Kim of Handong University, our center partner, and a github link related to this is also attached.