In this section, you can configure the voice your AI agent will use when speaking with callers. At the moment, Salpre AI integrates with ElevenLabs for text-to-speech (TTS), but we are continuously adding more providers so you can choose from a wide range of voices and styles.
🔹 Why Voice Selection Matters
The voice of your AI agent is the first thing callers experience. It sets the tone of trust, professionalism, and clarity. Choosing the right voice can significantly improve how comfortable and engaged your customers feel.
🔹 Why Female Voices Work Best for Calls
Through testing and customer feedback, we have found that female voices are often more effective in phone conversations:
- Better Clarity – Phone calls compress audio heavily (narrowband frequency, low bitrate). Female voices typically remain clearer and more understandable in these conditions.
- Pleasant Tone – Higher-pitched voices often sound warmer and more friendly in noisy call environments.
- Industry Standards – Many call centers and IVR systems worldwide default to female voices for smoother user experience.
Note: You can still select male voices if preferred, but for best performance on phone networks, we recommend female voices.
🔹 Language Selection
Selecting the correct language is critical for proper pronunciation and flow.
- If your agent speaks Turkish, select a Turkish voice to ensure natural pronunciation.
- For multilingual businesses, you can create multiple agents with different languages, or instruct your prompt to allow language switching.
- Wrong language settings may cause mispronunciation (e.g., English voice reading Turkish names incorrectly).
🔹 Accent & Pronunciation
Within a single language, accents make a big difference:
- English (US, UK, Australian, Indian, etc.) → Choose based on your customer base.
- Spanish (Spain vs. Latin America) → Select the version your customers expect.
- French (France vs. Canada) → Keep it regionally consistent.
Why it matters: Callers feel more comfortable and understood when the AI matches their accent and pronunciation style.
🔹 ElevenLabs Voice Possibilities
ElevenLabs offers advanced neural voice models with the following advantages:
- Ultra-realistic voices – Natural intonation and expressive delivery.
- Multilingual support – Dozens of languages with strong pronunciation accuracy.
- Custom voice cloning (for advanced users) – Create unique branded voices.
- Fine-tuning options – Adjust stability, clarity, and expressiveness for best results.
🔹 Future Voice Providers
While ElevenLabs is our current partner, we are expanding support to other TTS providers, such as:
- Google Cloud TTS (wide language coverage, fast latency)
- Azure Cognitive Services (enterprise-ready, stable)
- OpenAI Voice Models (direct voice-to-voice capabilities)
This ensures that in the future you can choose the voice provider that best matches your budget, latency requirements, and brand identity.
📝 Best Practices for Voice Settings
- Start with a female voice for clarity on mobile/landline calls.
- Match voice language to your customer base for natural flow.
- Select the right accent (e.g., UK vs. US English) to improve trust.
- For branding, consider custom voice cloning (premium feature with ElevenLabs).
- Test different voices with your team before finalizing — small differences can change how your customers perceive the AI.