Match Accent to Market
Choosing the right regional accent improves caller trust and comprehension:
| Market | Recommended accent |
|---|
| North America | US English |
| Europe | UK English |
| ANZ | Australian English |
| Mexico | Mexican Spanish |
| Spain | Castilian Spanish |
| Argentina | Argentine Spanish |
| Brazil | Brazilian Portuguese |
| Portugal | European Portuguese |
Always test with native speakers in your target market before deploying. Small accent mismatches can significantly affect caller perception.
Multilingual Agents
Enable language detection to let your agent respond in the caller’s language automatically.
Create language-specific knowledge bases using Genius:
- English Genius → English voice agent
- Spanish Genius → Spanish voice agent
- French Genius → French voice agent
This ensures culturally appropriate, accurately translated responses.
Always match the voice’s language to the agent’s conversation language. Using a mismatched voice/language pair causes pronunciation artifacts that callers notice immediately.
Advanced Controls (Cartesia)
Cartesia voices expose additional audio controls in the Presence section of the Agent Builder sidebar. These controls are only active when a Cartesia voice is assigned to the agent. If you are using an ElevenLabs voice, the controls are disabled with a prompt to switch to a Cartesia voice.
Speed
The Voice Speed slider controls how fast or slow the agent speaks.
| Setting | Effect |
|---|
| 0 | Slowest speech |
| 50 (default) | Normal speed |
| 100 | Fastest speech |
Start at the default (50) and make small adjustments. Even a shift of 10 points is noticeable. Test with Call Me to hear the difference in a real call. Volume (Coming Soon)
A Voice Volume slider is in development. It will allow you to increase or decrease the loudness of the agent’s speech independently of your phone system’s volume. Once released, the control will appear alongside the Speed slider in the Presence section.
Voice Vendor Comparison
| Feature | Cartesia | ElevenLabs |
|---|
| Strengths | Ultra-low latency, expressive speech with advanced audio controls | Wide accent and language coverage, rich tonal variety |
| Voice Selection | Professional voices with speed, emotion, and tuning options | Large catalog with diverse personalities |
| Best For | Real-time applications requiring low latency and expressiveness | Natural conversations with broad accent needs |
| Licensing | Subject to Cartesia terms | Subject to ElevenLabs terms |
Pronunciation Fixes
If your agent mispronounces names, brands, or industry terms:
- Phonetic spelling in prompts: write “Thought-lee” instead of “Thoughtly”
- Genius pronunciation guide: add entries to your Genius knowledge base with the correct spoken form
- Voice cloning: clone a voice that naturally handles your domain vocabulary
