Skip to content

ToucanTTS

A multilingual and controllable text-to-speech toolkit

Visit official site

What toucantts is

The toolkit is described as a multilingual and controllable text-to-speech system from the University of Stuttgart. It is built with plain Python and PyTorch for teaching, training and research use.

What you can do with it

  • Synthesize speech in multiple languages and voices
  • Set custom language and speaker embeddings
  • Fine-tune a pretrained model or train from scratch

Getting started

  • Clone the toolkit to a local machine
  • Create a virtual environment and install dependencies
  • Configure storage paths and pretrained models
  • Download pretrained models with the provided scripts

Pros & cons

✓ What we like

  • Built with plain Python and PyTorch
  • Detailed install and usage guides

! What to watch out for

  • Pretrained models are downloaded through scripts
  • A virtual environment and dependencies are part of setup

FAQ

Who built it?

The card names the NLP institute at the University of Stuttgart.

Can it be fine-tuned?

Yes, training from scratch and fine-tuning a pretrained model are both listed.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review