Skip to main content
Your AI assistant can speak with a built-in voice from the library or a custom cloned voice. Natural, realistic voices increase customer trust and engagement.

TTS provider

Select your Text-to-Speech provider in the assistant settings. This option is available in Pipeline and Dualplex modes. Available providers:
  • ElevenLabs — high-quality voices with a wide selection of languages and accents.
  • Cartesia — fast, low-latency synthesis, useful when response speed is a priority.
The TTS provider dropdown appears after you select a language for your assistant.

Choosing a voice

Each TTS provider has its own voice library. You can filter by gender, accent, or language depending on your provider. Browse the options in your assistant’s voice settings and preview voices before committing.

Cloning a voice

Voice cloning creates a custom voice from an audio sample — ideal for brand consistency, matching a company spokesperson, or building a personal connection with callers. Cloning is available in Pipeline and Dualplex modes. Requirements by provider:
ProviderRequirements
CartesiaSingle audio file, at least 10 seconds, one speaker, no background noise
ElevenLabsAudio samples totalling over 1 minute (max 5 minutes), one speaker, no background noise. Quality matters more than quantity.
Steps to clone a voice:
1

Start the cloning process

In your assistant’s voice settings, click Clone voice next to the voice selector.
2

Choose a provider

Select Cartesia or ElevenLabs based on your needs.
3

Set the language

Choose the language the cloned voice will speak.
4

Name your voice

Enter a name you’ll recognise in the dropdown later.
5

Upload audio

Record a new sample or upload an existing file that meets the provider’s requirements.
6

Wait for processing

Processing typically takes a few minutes. You’ll see the voice appear in the dropdown when it’s ready.
7

Select your new voice

Pick the cloned voice from the voice selector dropdown and save your assistant.

Best practices

  1. Use high-quality audio — clearer samples produce better clones.
  2. Record with steady delivery — natural tone, no abrupt pauses or changes in volume.
  3. Eliminate background noise — record in a quiet room or use a quality microphone.
  4. Get proper consent — only clone voices you have permission to use.
After selecting or cloning a voice, make a test call to confirm it sounds as expected before deploying to a live campaign or number.