Skip to content

Text-To-Speech plugins

The Text-To-Speech plugins allow you to connect Open Voice OS with different TTS servers, it could be Piper, Coqui, Amazon Polly, Mimic, etc... Each of these TTS providers will have different voices and languages.


The Text-To-Speech plugins are handled by the ovos_audio container.

The ovos_audio container comes with few pre-installed TTS plugins such as:

  • ovos-tts-plugin-polly is the Amazon Polly TTS server (authentication required)
  • ovos-tts-plugin-server allows you to reach an external TTS service

If the existing TTS plugins are not enough then you can install yours by following the same principle as for the skills by adding an audio.list file within the ~/ovos/config/ directory, this file acts as a Python requirements.txt file.

Plugins requirements

These plugins have to be compatible with the pip install method which requires a file.

When the ovos_audio container starts, it will look for this file and install the plugins defined in there.

ovos-tts-plugin-marytts==0.0.1a1 # Specific plugin version on PyPi
neon-tts-plugin-mozilla-remote # Latest plugin version on PyPi
git+ # Specific branch of a plugin on GitHub

The ovos_audio container must be restarted if a change occurs in the audio.list file.

docker restart ovos_audio
podman restart ovos_audio