---
title: "Agent Settings"
description: "Configure voice, AI providers, transcription, language, personality, and call behavior."
publishedAt: "2026-06-27"
modifiedAt: "2026-07-07"
category: "Agent Setup"
tags: "Voicetta, agent settings, voice, AI, transcription"
navOrder: "14"
canonical: "https://voicetta.com/docs/agent-settings"
---

# Agent Settings

Agent settings control how the agent sounds, listens, thinks, and behaves during calls.

Use this page after the greeting and core prompt are in place.

## Why it matters for your business

- **Save money:** choose model and voice settings that fit your quality and cost needs.
- **Work faster:** tune the voice pipeline without rebuilding the agent.
- **Standardize service:** keep tone, language, and behavior consistent across calls.
- **Delight every Guest:** a clear voice and reliable conversation flow feel more human and respectful.

## What you can configure

Depending on your workspace setup, settings may include:

- voice and text-to-speech provider,
- speech-to-text provider,
- LLM provider and model,
- language,
- personality and speaking style,
- call behavior,
- real-time or chained voice pipeline,
- provider API keys.

## Always-current models and voices

Model and voice lists are synced automatically with our AI partners — you always see the current catalog without waiting for an app update:

- New models appear in the pickers as soon as our platform supports them.
- ElevenLabs and Cartesia voice lists refresh directly from the providers.
- OpenAI Realtime voices (built-in pipeline) are discovered live from OpenAI, with a short description for each voice.
- Models labeled **"(voicetta)"** can run through the Voicetta subscription — no provider account needed.

## Who pays for a provider

Each provider in the pickers shows a billing badge:

| Badge | Meaning |
|-------|---------|
| **Your key** | You added your own API key — the provider bills you directly, model usage is free on the Voicetta side. |
| **Voicetta** | No key needed — usage runs through Voicetta and is billed to your wallet at the listed per-minute rate. |
| *(No API key)* | The provider needs a key that neither you nor Voicetta has — add one in Settings to unlock it. |

You can freely pick any provider with a **Voicetta** badge even without your own accounts.

## Step-by-step in the app

1. Open **Agent → Agent settings**.
2. Review the selected voice.
3. Review AI and transcription providers and their billing badges.
4. Check billing notes for platform-billed usage or BYOK usage.
5. Save changes.
6. Run a test call.

## Billing note

If you bring your own provider keys, that provider may bill you directly for model usage. Without your own keys, model usage is billed by Voicetta at the rates shown next to each model. Voicetta may still charge wallet-based platform, phone, or carrier fees depending on the call path.

## Best practices

- Test after every voice or model change.
- Use the simplest model that answers accurately.
- Match voice style to your brand.
- Keep business rules in prompts, not only in model selection.

## Related pages

- [API Keys and BYOK](/docs/api-keys-byok)
- [Billing, Wallet, and Pricing](/docs/billing-and-wallet)
- [Test Your Agent](/docs/test-your-agent)