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

# Voice Profiles

> Create unique voice identities for your AI agents with natural-sounding voices and custom voice cloning

## Overview

Voice is the identity of your AI agent. eigi.ai offers a diverse selection of natural-sounding voices and the ability to create custom voice profiles that perfectly represent your brand.

<Info>
  Your agent's voice creates the first impression. Choose a voice that matches
  your brand personality and resonates with your audience.
</Info>

***

## Voice Selection

### Pre-Built Voice Library

Access a curated library of professional voices:

<CardGroup cols={2}>
  <Card title="Natural Voices" icon="user">
    Human-like voices with natural intonation, pacing, and emotional expression
  </Card>

  <Card title="Multiple Languages" icon="globe">
    Support for various languages and regional accents
  </Card>

  <Card title="Voice Styles" icon="sliders">
    Professional, friendly, authoritative, warm, and more
  </Card>

  <Card title="Gender Options" icon="users">
    Male, female, and neutral voice options
  </Card>
</CardGroup>

### Choosing the Right Voice

Consider these factors when selecting a voice:

| Factor       | Consideration                                                                 |
| ------------ | ----------------------------------------------------------------------------- |
| **Industry** | Healthcare may need calm, reassuring voices; sales may prefer energetic tones |
| **Audience** | Match voice characteristics to your target demographic                        |
| **Brand**    | Align voice personality with your brand identity                              |
| **Use Case** | Support agents may need different voices than sales agents                    |

***

## Voice Cloning

Create a unique voice that's exclusively yours.

<Warning>
  Voice cloning requires proper consent and authorization. Only clone voices you
  have legal rights to use.
</Warning>

### How Voice Cloning Works

<Steps>
  <Step title="Record Sample">
    Provide a clear audio sample (typically 30 seconds to 3 minutes) of the
    voice you want to clone
  </Step>

  <Step title="AI Processing">
    Our system analyzes the voice characteristics—tone, pitch, cadence, and
    unique qualities
  </Step>

  <Step title="Voice Model Created">
    A custom voice model is generated that captures the essence of the original
    voice
  </Step>

  <Step title="Deploy to Agent">
    Assign the cloned voice to your agent for use in conversations
  </Step>
</Steps>

### Best Practices for Voice Samples

For optimal voice cloning results:

<AccordionGroup>
  <Accordion title="Recording Quality" icon="microphone">
    * Use a quiet environment with minimal background noise - Record in a studio
      or well-insulated room - Use a quality microphone (USB or XLR) - Maintain
      consistent distance from the microphone
  </Accordion>

  <Accordion title="Content Guidelines" icon="file-audio">
    * Speak naturally and clearly - Include varied sentences (questions,
      statements, exclamations) - Cover different emotions and tones - Avoid
      reading monotonously—speak as you would in conversation
  </Accordion>

  <Accordion title="Technical Requirements" icon="settings">
    * Audio format: WAV or MP3 - Sample rate: 44.1kHz or higher - Bit depth:
      16-bit or higher - Length: 30 seconds minimum, 3+ minutes recommended
  </Accordion>
</AccordionGroup>

***

## Voice Customization

### Adjusting Voice Parameters

Fine-tune your agent's voice for the perfect delivery:

| Parameter      | Description                                   |
| -------------- | --------------------------------------------- |
| **Speed**      | Control how fast or slow the agent speaks     |
| **Pitch**      | Adjust the voice pitch higher or lower        |
| **Stability**  | Control voice consistency vs. expressiveness  |
| **Similarity** | How closely to match the original voice model |

### Emotional Expression

Natural conversations involve emotional variation. Configure how your agent expresses:

* **Enthusiasm**: For exciting announcements or positive responses
* **Empathy**: For sensitive customer service situations
* **Confidence**: For authoritative information delivery
* **Warmth**: For friendly, approachable interactions

***

## Voice Providers

eigi.ai integrates with leading voice synthesis technologies:

<CardGroup cols={2}>
  <Card title="High-Fidelity Voices" icon="star">
    Ultra-realistic voices with nuanced expression and natural prosody
  </Card>

  <Card title="Low-Latency Synthesis" icon="bolt">
    Fast voice generation for real-time conversations without noticeable delay
  </Card>

  <Card title="Multilingual Support" icon="language">
    Voices that support multiple languages and code-switching
  </Card>

  <Card title="Custom Training" icon="graduation-cap">
    Train custom voice models for unique requirements
  </Card>
</CardGroup>

***

## Use Cases

### Brand Voice Consistency

Create a signature voice that represents your brand across all touchpoints:

* Phone support
* Website widget
* Mobile app
* Video content

### Executive Voice Cloning

Clone leadership voices for:

* Personalized customer outreach
* Training and onboarding content
* Announcement messages

### Localization

Deploy agents with native-sounding voices in different regions:

* American English for US markets
* British English for UK markets
* Spanish for Latin American markets
* And many more...

***

## Voice Testing

Before deploying, test your voice configuration:

<Steps>
  <Step title="Preview">
    Listen to sample phrases spoken by your configured voice
  </Step>

  <Step title="Test Conversation">
    Have a test conversation with your agent
  </Step>

  <Step title="Gather Feedback">
    Have team members evaluate the voice quality and fit
  </Step>

  <Step title="Iterate">Adjust parameters based on feedback</Step>
</Steps>

***

## Best Practices

<Tip>
  **Consistency**: Use the same voice across all agents in a project for brand
  cohesion.
</Tip>

<Tip>
  **Speed Matters**: Slightly slower speeds often work better for complex
  information; faster for casual conversations.
</Tip>

<Tip>
  **Test with Real Content**: Use actual prompts and responses when evaluating
  voices, not generic test phrases.
</Tip>

<Note>
  Voice technology is continuously improving. Check back regularly for new
  voices and enhanced capabilities.
</Note>

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="Voice sounds robotic">
    * Try a different voice from the library - Adjust stability settings for
      more natural variation - Check if your prompts are causing unnatural
      phrasing
  </Accordion>

  <Accordion title="Voice cloning quality issues">
    * Provide a longer, cleaner audio sample - Ensure consistent audio quality
      throughout the sample - Remove any background noise from recordings
  </Accordion>

  <Accordion title="Voice latency">
    * Use low-latency voice options for real-time conversations - Check your
      network connection - Consider regional server selection
  </Accordion>
</AccordionGroup>