Speech (Voice In/Out)

Your agent can accept voice input and reply with audio. Both can be enabled from Model Settings.

Where to find speech settings

Open “Model Settings” on the Agents page and look for the Speech section.

What you can enable

• Speech to Text: let users record via microphone.
• Text to Speech: play the agent’s reply as audio.
• Autoplay: automatically start playback after a reply.
• Voice: choose a voice when the provider supports it.
• ElevenLabs: use workspace‑level voices or your personal key (on paid plans).

Steps

1. Open the dashboard (opens in a new tab) and select your agent.
2. Click "Model Settings".
3. Under "Speech", toggle the features you want:
   • Enable “Speech to Text” for microphone input.
   • Enable “Text to Speech” to hear replies.
   • Turn on “Autoplay” if you want audio to start automatically.
   • Choose a Speech model and a Voice when available.
4. If you use ElevenLabs on a paid plan, you may enable “Use my ElevenLabs key” and add your API key.
5. Close the popover. Your changes save automatically.

Notes

• On free plans, ElevenLabs appears as an upgrade option.
• OpenAI TTS and other providers are supported where available.
• Public agent pages respect your speech settings.