> ## Documentation Index
> Fetch the complete documentation index at: https://help.thedouble.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Voice Overview

Voice lets your AI Double speak in real-time conversations with visitors — using either a cloned version of your voice or a preset voice from the library.

Go to **Double → Voice** to get started.

### **What Voice does**

* Clones your voice from an uploaded or recorded audio sample
* Powers real-time voice conversations with visitors on any profile or embedded chatbot
* Works alongside chat — both channels use the same knowledge, Persona, and behavior settings
* Supports 30+ languages with automatic detection

### **Voice Types**

There are two voice tiers. Both work for real-time conversations — the difference is quality and source.

|                | **HQ Voice**                               | **Light Voice**                        |
| :------------- | :----------------------------------------- | :------------------------------------- |
| **What it is** | High-quality model — cloned or preset      | Lightweight model — preset only        |
| **Best for**   | Most natural, human-like output            | Faster responses, lower resource usage |
| **Examples**   | Rachel, Ryan, Sakshi, Vijay, Linda, Thomas | Heart, Bella                           |

**Note:** HQ voices are marked with an **HQ** badge in the Voice Library. Light voices are marked **LITE**.

### **Choosing a Voice**

You have three options:

1. **Clone your own voice** — record or upload a sample and create a voice that sounds like you
2. **Pick a preset HQ voice** — choose from the Voice Library (Rachel, Ryan, Sakshi, Vijay, Linda, Thomas, and more)
3. **Pick a preset Light voice** — faster option for lower-latency use cases (Heart, Bella)

Click **Play** next to any preset to preview it before selecting.

### **Default Voice Language**

Set the default language visitors start with when opening a voice chat. This is the language hint sent to speech recognition.

Go to **Voice → Language** and select your preferred default. English (US) is set by default.

Visitors see the language flag preselected in voice chat. Speech recognition will prefer that language when voice chat starts. Visitors can still switch languages manually during a conversation.

### **Expressive Mode**

Toggle **Expressive Mode** on to add subtle, human-like expression to your Double's voice — natural pauses, hesitation, or laughter when it fits the conversation. Makes your Double sound less robotic and more like you.

Expressive Mode is off by default. Toggle it on from the Voice dashboard.

### **Multilingual Support**

Your Double speaks 30+ languages — automatically.

When a visitor starts a conversation, your Double detects their language and responds in the same language. No setup needed.

* **Auto detect** (default) — your Double identifies the language from the visitor's message and replies in kind
* **Manual selection** — visitors can click the country language flag icon in the chat input to pick their preferred language

Supported languages: Arabic, Burmese, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Indonesian, Italian, Japanese, Khmer, Korean, Lao, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Tagalog, Thai, Turkish, Vietnamese

**What to know**

* Multilingual support works for both text chat and voice calls
* Visitors can switch languages mid-conversation
* Voice responses are delivered in the visitor's selected or detected language
* The default language set in Voice settings is the starting hint — visitors can override it

### **Pronunciation**

Control how your Double's voice agent pronounces your name and company — without changing what visitors see in the chat transcript.

Go to **Voice → Pronunciation** to set it up.

1. Select your **Profile** from the dropdown
2. Enter your **First name** and **Last name** as they appear on your profile
3. Enter your **Company on Hero card**
4. In **First name pronunciation (IPA)**, type how your first name should sound (e.g. "Ma-axe")
5. In **Last name pronunciation (IPA)**, type how your last name should sound (e.g. "Pay-ain")
6. In **Company pronunciation (IPA)**, type how your company name should sound (e.g. gu:gəl)
7. Click **Generate pronunciation** to preview, or **Save pronunciation** to apply

**Note:** IPA fields control the voice only. The transcript always displays your name and company exactly as entered on your profile.

**Tips for a good voice clone**

* **Avoid noisy environments** — background sounds interfere with recording quality
* **Use a good microphone** — a studio mic works best; on phones, use Voice Memos (iOS) or Dolby On (Android)
* **Finish sentences naturally** — avoid cutting off words or stopping mid-sentence

**Next steps:**

* Cloning Your Voice
* Choosing a Preset Voice
* Voice Quality and Troubleshooting