Can I set different TTS models/voices per agent?
✅Solved
Hey Krill, I’m trying to configure different TTS models (and/or voices) for different agents in OpenClaw.
What I’m trying to do
I have multiple agents (e.g. main, work, home) and I want:
• Agent home → ElevenLabs voice/model A
• Agent work → OpenAI TTS voice/model B
(or at least different voices/models while using the same provider)
What’s happening instead
Right now, TTS seems to be controlled globally by messages.tts, so all agents end up using the same TTS settings.
Is there an official way to set different TTS models/voices per agent?
What I’m trying to do
I have multiple agents (e.g. main, work, home) and I want:
• Agent home → ElevenLabs voice/model A
• Agent work → OpenAI TTS voice/model B
(or at least different voices/models while using the same provider)
What’s happening instead
Right now, TTS seems to be controlled globally by messages.tts, so all agents end up using the same TTS settings.
Is there an official way to set different TTS models/voices per agent?
Solution
Right now: no, there isn’t an “official” per-agent TTS config block. The config schema only has global
Docs: https://docs.openclaw.ai/tts (see “Config” + “Model-driven overrides”)
#### 1) Different voices/models per agent (same provider)
Use the built-in TTS directive tags in the agent’s replies:
- OpenAI voice/model per reply:
- ElevenLabs voice/model per reply:
This is the intended mechanism for “per reply” overrides, and you can make it “per agent” by putting a rule in each agent’s workspace prompt files (SOUL/AGENTS) like “Always include this
#### 2) Different providers per agent (OpenAI vs ElevenLabs)
Also possible, but you must explicitly opt in:
messages.tts/tts provider …Docs: https://docs.openclaw.ai/tts (see “Config” + “Model-driven overrides”)
What you can do today (supported)
#### 1) Different voices/models per agent (same provider)
Use the built-in TTS directive tags in the agent’s replies:
- OpenAI voice/model per reply:
[[tts:voice=alloy model=gpt-4o-mini-tts]]- ElevenLabs voice/model per reply:
[[tts:voiceId=... model=eleven_v3]]This is the intended mechanism for “per reply” overrides, and you can make it “per agent” by putting a rule in each agent’s workspace prompt files (SOUL/AGENTS) like “Always include this
[[tts:...]]#### 2) Different providers per agent (OpenAI vs ElevenLabs)
Also possible, but you must explicitly opt in:
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
131,281Members
Resources
Recent Announcements
Similar Threads
Was this page helpful?
