Coqui TTS
Text-to-speech generation with multilingual support
45.1k|stable|v0.22.0
from TTS.api import TTS
tts = TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC", gpu=True)
tts.tts_to_file(text="Hello world", file_path="output.wav")
1100+ pretrained models — Covers over 1100 languages with ready-to-use speech synthesis models.
Multi-speaker TTS — Generate speech in different speaker voices using speaker embeddings.
Voice cloning — Clone voices using short audio samples with models like XTTS.
Model training pipeline — Train custom TTS models or fine-tune existing models on new datasets.
Multiple vocoder implementations — Choose from MelGAN, ParallelWaveGAN, HiFiGAN and other vocoder architectures.
Battle-tested production library with models like XTTS that handle 16 languages and stream with sub-200ms latency. Includes multiple spectrogram and end-to-end architectures (Tacotron, Glow-TTS, VITS, Bark) plus comprehensive training utilities for dataset curation and model development.
Python 3.8+, PyTorch; GPU recommended for inference
Source on GitHub
pavilion