coqui-ai/TTS

Project overview

A deep learning toolkit for Text-to-Speech generation that provides pretrained models in over 1100 languages, voice cloning, and model training tools. It is intended for developers, AI engineers, and researchers building speech synthesis applications.

Project type

Model Runtime · Audio & Speech

Use cases

Audio & Speech

Deployment

Refer to project documentation

License

MPL-2.0

Best for

• Developers and AI engineers needing a self-hosted, local-run library for synthesizing speech or building custom TTS models.
• Researchers requiring access to various Text2Spec and vocoder architectures alongside a Trainer API for fine-tuning.

Key capabilities

• Synthesizes speech from text using pretrained or custom models, with support for over 1100 languages.
• Provides tools and a Trainer API to train new models and fine-tune existing Text-to-Speech models efficiently.
• Clones voices using speaker embeddings or reference audio, enabling unconstrained voice cloning with compatible models like XTTS and Bark.
• Converts the voice from a source audio file to match a target voice using models like FreeVC.
• Provides a built-in server to run TTS models and synthesize speech over HTTP, exposed via a web GUI.
• Offers utilities under 'dataset_analysis' for curating Text2Speech datasets.
• Includes various Text2Spec models like Tacotron, Glow-TTS, FastPitch, and more.
• Includes various vocoder models like MelGAN, HiFiGAN, WaveRNN, and more to convert spectrograms to audio.

Limitations and risks

• The library is tested primarily on Ubuntu 18.04. Running it on Windows requires following manual setup instructions.

Getting started

• Installation difficulty is rated as easy. It is available on PyPI for simple installation via pip and provides a ready-to-use Docker image.

Evidence and sources

• GitHub project description: 🐸💬 - a deep learning toolkit for Text-to-Speech, battle-tested in research and production
• README: 🐸TTS is a library for advanced Text-to-Speech generation.
• README: tts.tts_to_file(text="Hello world!", speaker_wav="my/cloning/audio.wav", language="en", file_path="output.wav")
• README: - High-performance Deep Learning models for Text2Speech tasks. - Text2Spec models (Tacotron, Tacotron2, Glow-TTS, SpeedySpeech). - Speaker Encoder to compute speaker embeddings ef…
• README: tts = TTS(model_name="tts_models/multilingual/multi-dataset/your_tts", progress_bar=False).to(device) tts.tts_to_file("This is voice cloning.", speaker_wav="my/cloning/audio.wav",…

Reviewed at: 2026-07-17T07:23:38Z