Deep Learning Library
10.1k 2026-05-05
mozilla/TTS
A deep learning library for advanced, high-quality, and efficient Text-to-Speech (TTS) synthesis, supporting multiple languages and models.
Core Features
High-performance Deep Learning models for Text2Speech (e.g., Tacotron, Glow-TTS, Vocoders).
Support for multi-speaker and efficient multi-GPU training.
Pretrained models available and used in over 20 languages.
Ability to convert PyTorch models to TensorFlow 2.0 and TFLite for inference.
Tools for dataset quality measurement and analysis.
Quick Start
tts --text "Hello world" --model_name "tts_models/en/ljspeech/tacotron2-DDC" --out_path output.wavDetailed Introduction
Mozilla TTS is an advanced open-source deep learning library designed for Text-to-Speech generation. Built on cutting-edge research, it offers an optimal balance of training ease, speed, and output quality. The library includes a variety of state-of-the-art models, supports over 20 languages, and provides tools for dataset curation and model deployment across different frameworks like PyTorch, TensorFlow, and TFLite. It's a comprehensive solution for researchers and developers building custom TTS applications.