Voices and cloning
Your voice. Your tone. Your phone reception.
Choose a voice that fits your business or use a clone of your own voice. With ElevenLabs and compatible models, give your assistant a more expressive way of speaking.
Create my assistant« Hello and welcome. How can I help you today? »
- Warm welcome
- Reassuring tone
- Unhurried pace
A voice that fits your business
A warm welcome, a calm explanation or a more energetic reply: voice choice shapes the caller’s experience. ReplyBot lets each agent use a speech synthesis provider, including ElevenLabs, independently of the model that prepares replies.
Clone your own voice
Create your voice clone in ElevenLabs using your recordings, under the terms of its plan. To use it in ReplyBot, the voice must be accessible to the ElevenLabs account connected to the integration. Then enter its voice ID in the agent’s voice settings. Cloning takes place with the provider; our team can help you set it up for calls.
Intonation and vocal expressions
With a compatible ElevenLabs model, text cues can guide intonation or add an expression such as a light laugh or a whisper. Define the desired tone in the agent’s instructions and test the result: audio tags and their delivery vary by model and voice.
Adjust, listen, then go live
ElevenLabs stability and similarity settings help adjust consistency and closeness to the reference voice, depending on the model. Test names, numbers, interruptions and agent handoffs in the browser and then over the phone. The network and handset also affect what callers hear.
From setup to your first call
- 01
Choose or clone
Select a voice or prepare your clone in ElevenLabs.
- 02
Assign to an agent
Use a voice accessible to the integration and set the tone.
- 03
Test a call
Listen to your business phrases before going live.
Your questions answered
Does cloning happen inside ReplyBot?
No. Create the clone in ElevenLabs, then use its voice ID in ReplyBot, subject to access from the connected account and compatibility with the selected model. The provider’s terms and any fees apply.
Can I add emotions to every voice?
Expressions and audio tags require a compatible speech synthesis model. They are not universal and will not always sound identical. Instructions guide the tone; you still need to test your own phrases.
Can each sub-agent have its own voice?
Yes. Each Pipecat agent can use its own voice and speech synthesis provider. Keep a shared identity or distinguish reception, booking and support.
Which voices can I clone?
Use your own voice and complete the provider’s verification steps. For any other voice, ensure you hold the required rights and permissions and follow the service’s rules. A custom voice should not hide the fact that the caller is speaking with AI.