• All Collections
  • Clone a Voice

How to Create Custom Voices from Text Descriptions

Kukarella's text-to-voice generation feature allows you to create custom voices by simply describing what you want in plain text. Instead of browsing through our library of 1,800+ voices, you can describe the exact voice characteristics you need and our AI will generate a custom voice for you.

Key Advantage: Preview Before Creating! Unlike audio-cloned voices, text-generated voices offer a preview option, allowing you to test and refine before committing to creation.


When to Use This Feature

This feature is perfect when you:

  • Have a specific voice in mind that's hard to find in our voice library
  • Want to create character-like voices for storytelling or entertainment
  • Need a voice with very specific characteristics (accent, tone, pace, etc.)
  • Want to save time instead of listening through multiple voice samples
  • Want to preview and test before using a clone slot

<iframe width="560" height="315" src="https://www.youtube.com/embed/hThBMxRoPTI?si=6YIh24B22Lda_E9D" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>

Step-by-Step Instructions

1. Access the Feature

2. Describe Your Voice

Write a detailed description of the voice you want. Include characteristics such as:

  • Gender: Male, female, or non-binary
  • Age: Young, middle-aged, elderly
  • Tone: Deep, high-pitched, gravelly, smooth, nasal
  • Pace: Fast, slow, measured, lazy
  • Style: Energetic, calm, dramatic, casual
  • Accent or regional characteristics
  • Unique qualities: Slurred speech, frequent pauses, trailing off, etc.

3. Use Example Prompts (Optional)

If you're unsure how to describe your voice, you can:

  • Choose from provided example descriptions
  • Use them as references for writing your own prompts
  • Modify existing examples to better match your needs

4. Preview the Voice

  • Click the generate voice preview button
  • Wait 10-20 seconds for the AI to process your description
  • Listen to the generated voice sample
  • This preview does NOT use a clone slot

5. Refine if Needed

  • If the voice isn't quite right, edit your description
  • Add more specific details or adjust existing characteristics
  • Regenerate the preview as many times as needed
  • Only create the voice once you're satisfied

6. Create the Voice

  • Once you're happy with the preview, save it to your library
  • Important: After creation, the voice cannot be deleted or modified
  • The voice will use one of your clone slots

Writing Effective Voice Descriptions

Good Description Examples:

Example 1: "Deep, gravelly male voice with slow, lazy delivery and frequent 'doh' grunts. Nasal quality with slightly slurred speech. Pitch drops at sentence ends and often trails off mid-thought. British accent"

Example 2: "High-pitched, energetic female voice with a slight Southern accent and cheerful tone."

Example 3: "Elderly male voice with wise, measured delivery and slight tremor. French Parisian accent"

Tips for Better Results:

  • Be specific: Include multiple characteristics rather than just "male voice"
  • Use descriptive adjectives: Gravelly, smooth, nasal, breathy, etc.
  • Mention speech patterns: Fast delivery, pauses, trailing off, emphasis styles
  • Include personality traits: Cheerful, serious, mysterious, friendly
  • Describe language and accent

Character-Inspired Voices

You can create voices inspired by characters from:

  • Cartoons: Describe characteristics without using copyrighted names
  • Movies: Focus on vocal qualities rather than actor names
  • Fictional characters: Emphasize the voice traits that make them distinctive

Example: Instead of "SpongeBob voice," describe "high-pitched, extremely enthusiastic voice with childlike energy and rapid, excited delivery."


Understanding Voice Capabilities & Limitations

Multilingual Support

Your text-generated voice will be multilingual, capable of speaking in approximately 35 different languages and accents.

Language Detection

Important: Text-generated voices (like audio-cloned voices) use the first few words of your text to detect which language to speak.

Best practices for multilingual text:

  • Start with clear, unmistakable words in your target language
  • Avoid starting with ambiguous words that could be mistaken for another language
  • Example: For German text, start with clearly German words like "Der Botanische Garten" rather than words like "was" that could be interpreted as English

Tip: If you're having trouble with language detection, consider using our multilingual OpenAI-powered voices (marked with flame and masks icon) instead. They handle language switching more reliably and don't use clone slots.

Voice Management

Important: Once created, voices cannot be deleted or modified. This is a technical limitation of the voice cloning technology we use.

Before creating a voice:

  • Use the preview feature extensively to test
  • Ensure you're satisfied with all characteristics
  • Remember that each account has a limited number of voice clone slots

Troubleshooting

If the voice doesn't match your expectations:

  • Add more detail to your description
  • Emphasize specific characteristics that are most important
  • Try different adjectives to describe the same quality
  • Break down complex requests into simpler, more specific descriptions
  • Use the preview feature to test variations before creating

Best Practices

āœ… DO:

  • Start simple and add details gradually
  • Test multiple variations using the preview feature
  • Save successful descriptions for future reference
  • Experiment with different combinations of characteristics
  • Preview thoroughly before creating to save clone slots

āŒ DON'T:

  • Rush to create without previewing extensively
  • Use vague descriptions like "nice voice" or "good voice"
  • Expect to delete or modify after creation
  • Forget to test language detection with your actual text

Need Help?

If you're having trouble creating the voice you want:

  • Try our example prompts as starting points
  • Experiment with different descriptive words
  • Use the preview feature to test variations
  • Contact our support team for assistance at support@kukarella.com

Share Your Experience

We'd love to hear about your experience with this feature! Let us know:

  • What types of voices you've created
  • How well the feature works for your needs
  • Suggestions for improvements
  • Any creative uses you've discovered

Your feedback helps us make Kukarella better for everyone.


Remember: The preview feature is your friend - use it extensively before committing to voice creation!