XTTS Speaker Forge

A Gradio interface for Exploring, Creating, Editing, Mixing, Importing and Exporting Speakers for the Coqui XTTS model.

Run using Google Colab

• Google Colab Notebook

Run locally (Linux / Mac)

1. Clone the repo git clone https://github.com/ichabodcole/xtts-speaker-forge.git
2. CD into the newly created directory cd xtts-speaker-forge
3. Create virtual python environment via venv, conda, etc (optional, but highly recommended)
4. Install the packages pip install -q -r ./requirements.txt
5. Run the app bash run.sh
6. Click open the gradio app at localhost:5003

Run locally (Windows)

Not yet supported or tested, though it may work under WSL :D

Package install Errors

If you encounter package install errors during step #4 of "Run locally" try the alternative installation below (you only need to try this if the app will not actually run).

!pip install --use-deprecated=legacy-resolver -q -r ./requirements.txt
!pip install -q typing_extensions==4.8 numpy==1.26.2

Screenshots

• 
• 
• 
• 
• 
• 
• 

Roadmap

• Add language selection support for speaker audio generation
• Add "Edit" mode to allow renaming and deleting speakers, and adding additional speaker meta data.
• Audio file caching to quickly playback recently generated audio files.
• Allow editing model file paths in Gradio interface (if feasible)
• Allow direct upload / import of speaker files
• Explore adding different voice mixing methods.
• Ability to filter speakers based on metadata created in Edit view in Explore view.