Custom Voice Cloning: Using Your Own Cloned Voice

Overview

This page covers using a custom or cloned voice from a supported voice provider on an assistant. It's a different mechanism from the Custom Voice (TTS) page, which covers a fully custom webhook-based TTS integration rather than a cloned voice from an already-supported provider. See also Voice Fallback Plans for the per-provider settings reference.

Set the voice.provider and voice.voiceId fields directly in an assistant's configuration, either at creation or via a PATCH request to an existing assistant:

curl -X PATCH "https://api.sulus.ai/assistant/ASSISTANT_ID" \
  -H "Authorization: Bearer $SULUS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "voice": {
      "provider": "PROVIDER",
      "voiceId": "VOICE_ID"
    }
  }'

Or the same fields set directly in a JSON assistant configuration:

{
  "voice": {
    "provider": "deepgram",
    "voiceId": "your-voice-id"
  }
}

If you have a custom or cloned voice, or an enterprise account with a specific provider, add your own provider API key under the Integrations tab in the dashboard. Once validated, you won't be billed through the platform for that provider's usage — you'll be billed directly by the provider instead. Any voice ID from your own provider account can then be used in the assistant's voice.voiceId field.

To find the voice ID for a cloned voice already registered in your account:

1. Go to the dashboard and open Resources
2. Open the Voice Library
3. Find or search for your voice
4. Copy the voice ID
5. On the assistant, open the Voice card and paste the ID into Custom Voice ID

The following voice providers are supported for custom and cloned voices, many of which offer their own voice cloning capability:

• Sulus Voices
• ElevenLabs
• PlayHT
• Azure
• OpenAI
• Cartesia
• LMNT
• Minimax
• RimeAI
• Deepgram
• Inworld

The actual voice-cloning process — recording voice samples and training the clone — happens on the underlying provider's own platform, not within Sulus. Sulus only references the voice ID once the clone already exists in your provider account; it does not host or perform the cloning itself.

See the Custom Voice (TTS) page for the fully custom, webhook-based TTS integration mechanism — a different approach from using a cloned voice with an already-supported provider. See Voice Fallback Plans for the per-provider settings reference, including how to configure a fallback voice.