Add text to speech to Open WebUI

text-to-speech-open-webui.md

Add text to speech to Open WebUI

Clone the repository

https://github.com/matatonic/openedai-speech

Download voices

Download to directory voices, the language voices to use from:

https://huggingface.co/rhasspy/piper-voices/tree/main

Must download the .onnx and the .json files

Enable the added voices

Edit the config to enable the added voices. For every added voice, must supply and id into the config/voice_to_speaker.yaml file. For example, I added portuguese voice. For that, I downloaded:

pt_BR-faber-medium.onnx pt_BR-faber-medium.onnx.json

Then I copied those files to the voices directory. Then, I added to the config/voice_to_speaker.yaml file, the following key:

tts-1
  faber:
    model: voices/pt_BR-faber-medium.onnx
    speaker: # default speaker

Anything else at the file, remained the same.

After that, if the docker container is alredy running, must be restared

sudo docker restart <id_of_container>

Configure Open WebUI to use the new voice

Go to: Admin Panel -> Settings -> Voice

Once there, set the parameters to use the required voice: