mozilla/TTS

Project overview

Mozilla TTS is a deep learning text-to-speech toolkit that balances ease-of-training, speed, and quality. It provides pretrained models, dataset curation tools, multi-GPU training, and support for 20+ languages for developers and researchers building speech synthesis products.

Project type

Model Runtime · Data Processing · Audio & Speech

Deployment

Refer to project documentation

License

MPL-2.0

Best for

• Developers, AI engineers, and researchers building text-to-speech products who need deep learning models with multi-GPU training and dataset curation tools.

Key capabilities

• High performance deep learning models for Text2Speech tasks including Tacotron, Tacotron2, Glow-TTS, and SpeedySpeech.
• Vocoder models for audio generation including MelGAN, Multiband-MelGAN, GAN-TTS, ParallelWaveGAN, WaveGrad, and WaveRNN.
• Computes speaker embeddings efficiently to enable multi-speaker TTS.
• Provides a demo server for model testing.
• Ability to convert PyTorch models to Tensorflow 2.0 and TFLite for inference.
• Tools to curate Text2Speech datasets under dataset_analysis and a generic dataloader for custom datasets.
• Fast and efficient model training with multi-GPU support.
• Provides detailed training logs on the console and Tensorboard.

Limitations and risks

• The project requires a Python environment that is greater than or equal to 3.6 and less than 3.9.

Getting started

• Install the library from PyPI using pip, list available released models via the command line, and synthesize speech by providing text, model name, and an output path.

Evidence and sources

• README: TTS is a library for advanced Text-to-Speech generation. It's built on the latest research, was designed to achieve the best trade-off among ease-of-training, speed and quality.
• README: If you are only interested in [synthesizing speech](https://github.com/mozilla/TTS/tree/dev#example-synthesizing-speech-on-terminal-using-the-released-models) with the released TT…
• README: - High performance Deep Learning models for Text2Speech tasks. - Text2Spec models (Tacotron, Tacotron2, Glow-TTS, SpeedySpeech). - Speaker Encoder to compute speaker embeddings ef…
• Release: TTS v0.0.9 (first release)
• README: After the installation, TTS provides a CLI interface for synthesizing speech using pre-trained models. You can either use your own model or the release models under the TTS projec…

Reviewed at: 2026-07-18T03:35:34Z