Text Guided Text to Speech Generation

I converted the model to half precision and updated it as a cog so it can be commercialized on replicate more easily.

How to use

sudo cog predict -i prompt="hi hows it going" -i voice="A Well spoken english male clear voice no background noise"

example audio result

Dev setup

virtualenv .env
source .env/bin/activate
pip install -r requirements.txt

How to train

See original parler paper.

Clone the model

mkdir models
cd models
git clone [email protected]:spaces/parler-tts/parler_tts_mini

How to convert to half precision

Uncomment code in predict.py to do that, run it and then copy missing files over from the old full precision model folder.

How to convert to cog

I did that/thats what this repo is.

see predict.py

How to deploy

cog push

How to run tests

pytest .

How to run lint

flake8 predict.py

Please help me!!!

• Use a efficient output format not wav
• Support for more expressive and emotive voices
• Support for more languages
• Support for more voices
• Support for more accents

more cool audio stuff we should do somewhere

• Eleven labs style voice clone.
• Voice style transfer
• SunoAI - Inpainting of audio/generating anything audio
• Music style transfer

Let me know if these are of interest or if these have been done please link!

Plugs and sponsors For AI products.

See text to speech models on Text-generator.io https://text-generator.io/

AI Chat characters https://netwrck.com https://netwrck.com

AI Art Generation https://aiart-generator.io https://aiart-generator.io