Text-To-Speech Voices

Backyard uses the Piper text-to-speech model to enable real-time voice generation for your AI Characters.

Setup #

To enable text-to-speech for your Characters, head to the settings page and scroll down to the "text-to-speech" section.

On Desktop, text-to-speech runs locally (and offline) on your computer's CPU. It works out of the box on Windows and Mac without any additional setup.

On Web, text-to-speech runs on our servers and streams the audio to your browser. This requires remote processing, but we never log or store any of your data.

Customization #

• Presets: select from a catalog of 30+ unique voices for your Character. We will support more voices in the future.
• Speech Rate: control how fast the AI Character speaks. You can adjust the rate from 0.5x to 2x.
• Auto-Play: when enabled, your AI Charater will automatically start speaking after each response completes generation. When disabled, you will need to click the play button to hear the speech.
• Input Filter: allows you to selectively filter out certain sequences of words from the speech output. This is useful if you use a particular format for roleplay chatting. For example, if you use asterisks to denote actions, you can filter them out of the speech output.