mathieutrudeau / Fast-TTS

API that uses Tortoise and RVC to speed up text-to-speech generation.

Fast-TTS

Fast-TTS is a module that offers a simple api in order to generate a speech from text.

Requirements

• Python 3.9
• CUDA enabled GPU with PyTorch installed.

How to add Voices

The following steps show how to add voices.

• Create a folder with the speaker name '/MyVoice'. Here 'MyVoice' is the name of the speaker that you wish to add.
• Find 1-3 audio samples of the speaker, those samples should be ~1min in length and be .wav files. The clearer the voice, the better the generated audio will be.
• Add those audio sampples to the speaker folder that was previously created.
• Add the .index and .pth files for the speaker model. Name them 'MyVoice.index' and 'MyVoice.pth'.
• You can either get the .index and .pth files from trained RVC models, or you can train your own model.