Coqui TTS

Coqui TTS is an open-source deep learning toolkit for text-to-speech synthesis, originally developed by Coqui AI and now maintained by the community. It provides a unified interface to multiple neural TTS architectures including VITS, Tacotron2, GlowTTS, FastSpeech2, and Bark, with a built-in training pipeline for fine-tuning on custom datasets. The XTTS v2 model enables zero-shot voice cloning from a short reference clip and cross-lingual synthesis across 17+ languages.

The honest picture: Coqui AI the company shut down in 2024, so the project survives purely on community maintenance — updates are slower than commercial alternatives. Voice quality varies significantly by model, and running the larger models needs a decent GPU. But for teams that need full control, offline inference, or zero vendor lock-in, it remains the most battle-tested open TTS framework available.

pip install TTS, then run tts --text "Hello" --model_name tts_models/en/ljspeech/vits or use the Python API. Models download automatically on first use.