TensorSpeech/TensorFlowTTS
A TensorFlow 2 library for real-time text-to-speech synthesis using state-of-the-art models including Tacotron-2, FastSpeech, FastSpeech2, and neural vocoders.
Velocity · 7d
+1.8
★ / day
Trend
→steady
star history
TensorFlowTTS provides real-time speech synthesis implementations of deep learning models such as Tacotron-2, FastSpeech, and FastSpeech2, along with neural vocoders including Melgan and Multiband-Melgan. The library supports multi-speaker and multilingual TTS for English, French, Korean, Chinese, and German, with optimization via quantization and pruning for faster-than-real-time inference. It offers TensorFlow Lite export for mobile and embedded deployment.