Skip to main content
Bolly can speak its responses aloud using ElevenLabs text-to-speech. Voice mode is optional and requires an ElevenLabs API key.

​
Setup

  1. Add your ElevenLabs API key through Bolly’s settings in the browser
  2. Enable voice mode for your instance

​
Choosing a voice

Bolly can use any ElevenLabs voice. You can change the voice through:
  • The set_voice tool — ask Bolly to “change your voice” and provide a voice ID
  • The settings UI in the web interface
ElevenLabs offers 30+ preset voices with different characteristics (gender, accent, tone). You can also create custom voices on the ElevenLabs platform.

​
Per-instance voices

Each Bolly instance can have a different voice. The voice ID is stored in the instance configuration and takes precedence over any global default.

​
How it works

When voice mode is enabled:
  1. Bolly generates a text response as usual
  2. The text is sent to ElevenLabs TTS API
  3. An MP3 audio stream is returned
  4. The client plays the audio alongside the text response
Voice is a companion to text — you always get the written response too.