What It Does

Custom Voice Cloning lets you create AI voices that sound like real people—your CEO, brand spokesperson, or anyone else whose voice you want to use in marketing videos. Upload 1-5 audio samples, and ILLIXIS uses AI voice cloning to create a synthetic voice that matches the original speaker's tone, cadence, and characteristics.

Once created, your custom voice appears alongside preset voices in the video production workflow. Every video you produce can use your brand's unique voice instead of generic AI voices.

Requirements

Audio Sample Requirements

| Requirement | Specification |
|-------------|---------------|
| Format | MP3, WAV, or M4A |
| File size | Maximum 10MB per file |
| Number of files | 1-5 samples recommended |
| Total duration | 1-2 minutes of audio works best |
| Content | Natural speech only (no singing, music, or sound effects) |
| Quality | Clear recording without background noise or echo |

Best Audio Samples

For high-quality voice clones:

Use clean recordings - Studio quality or quiet room with no echo
Include varied speech - Different sentences, not repetitive phrases
Avoid reading style - Natural conversation sounds better than scripted reading
Multiple samples - 3-5 different recordings produce better results than one long file

Poor Audio Samples

Avoid these common issues:

Background music or ambient noise
Multiple speakers in the same recording
Phone recordings with compression artifacts
Echo from large rooms
Clipped or distorted audio

How to Create a Custom Voice

Step 1: Access Voice Library

Go to Settings in your main navigation
Select the Brand tab
Click the Voice Library tab

If you have no custom voices yet, you'll see an empty state with a "Create Your First Voice" button.

Step 2: Start Voice Creation

Click "Clone Voice" (or "Create Your First Voice" if this is your first).

You'll be taken to a full-page form for creating your voice clone.

Step 3: Enter Voice Details

Fill in the voice information:

| Field | Required | Description |
|-------|----------|-------------|
| Voice Name | Yes | Display name (e.g., "CEO Voice", "Brand Spokesperson") |
| Gender | No | Male, Female, or Neutral (helps categorize in dropdowns) |
| Language | No | Primary language: English, Spanish, Portuguese, French, German, or Italian |
| Description | No | Optional notes about when to use this voice |

Step 4: Upload Audio Samples

Click the upload zone or drag audio files directly onto it
Supported formats: MP3, WAV, M4A
Each file can be up to 10MB
Upload 1-5 samples for best results

As you add files, they appear in a list below the upload zone. Click the X icon on any file to remove it.

Step 5: Configure Noise Removal

By default, "Remove background noise" is enabled. This applies audio isolation to clean up your samples before cloning. Leave this enabled unless your recordings are already studio quality.

Step 6: Create the Voice Clone

Click "Create Voice Clone".

A processing overlay appears while ILLIXIS:

Uploads your audio samples to secure storage
Processes samples for voice cloning
Creates your custom voice profile
Generates a preview sample

This takes 30-60 seconds. Do not close the page during processing.

Step 7: Review Your Voice

After creation, you're redirected to the Voice Library. Your new voice appears in the "Your Custom Voices" section with:

Voice name and description
"Custom" badge
Gender badge
Play button to preview the voice
"Set Default" button (if not already default)
Delete button

Click the play button to hear a preview sample of your cloned voice.

Setting a Default Voice

When you have multiple custom voices, you can designate one as your default:

In the Voice Library tab, find the custom voice you want as default
Click "Set Default" on that voice card
The voice now shows a "Default" badge

The default custom voice is automatically selected when:

Creating new videos in Video Production
Using voice dropdowns throughout the platform
Your custom voice takes precedence over preset voices

To change your default:

Click "Set Default" on a different custom voice
The previous default is automatically cleared
The new voice becomes your default

Using Your Voice

In Video Production

Your custom voices appear in the voice selection dropdown when:

Creating a new video script
Editing storyboard voiceover settings
Regenerating individual scene voiceovers

To use a custom voice:

Go to Video Production
Create or open a video project
In the voice selection dropdown, look for the "Custom" section
Select your cloned voice
Generate the video—voiceovers use your custom voice

In Social Extensions

When creating video posts from articles or briefs:

Open the content piece in Social Hub
Click "Video Post" or "Create Video"
Select your custom voice from the dropdown
The generated video uses your voice for narration

Voice API Endpoint

For programmatic access, custom voices are included in the voice options API:

```
GET /content_hub/api/voices/
```

Returns both preset and custom voices for your tenant, with the type: "custom" field indicating cloned voices.

Managing Custom Voices

Maximum Voices

You can create up to 10 custom voices per account. If you've reached the limit, delete an existing voice before creating a new one.

Deleting a Voice

In the Voice Library, find the voice to delete
Click the trash icon on the voice card
Confirm the deletion

Deleting a voice:

Removes it from all dropdown menus
Preserves existing videos (they retain the audio already generated)
Cannot be undone (you would need to re-clone with new samples)

Voice Statuses

| Status | Meaning |
|--------|---------|
| Ready | Voice clone complete, available for use |
| Pending | Voice is being created (wait ~60 seconds) |
| Failed | Cloning failed—error message shown; try again with different samples |

Regenerating Preview

If the preview sample doesn't play or sounds wrong:

The system can regenerate a new preview
This creates a fresh sample using the same voice profile
Useful if the original preview had audio issues

Tips for Best Results

Recording Tips

Use the same microphone for all samples—mixing microphone types creates inconsistencies
Record in a quiet space with soft surfaces to reduce echo
Maintain consistent distance from the microphone (6-12 inches ideal)
Speak naturally as if having a conversation, not reading aloud
Include varied content—questions, statements, different emotions

Sample Content Ideas

Good content for voice samples:

Reading a blog post aloud naturally (not robotically)
Podcast recordings with that speaker isolated
Video testimonials or presentations
Casual conversation about a familiar topic

Poor content for voice samples:

Scripted marketing copy (sounds unnatural)
Reading numbers or lists
Whispered or exaggerated speech
Group conversations (isolate the target speaker)

Language Considerations

Voice clones work best when samples match the target language
English samples produce the best English output
Multilingual speakers should provide samples in each language if possible
Accent characteristics transfer to the clone (a British speaker produces British-accented output)

Quality vs. Quantity

3-5 good samples produce better results than 10 mediocre ones
1-2 minutes total is the sweet spot
Longer samples don't significantly improve quality
Focus on audio clarity over total duration

Troubleshooting

"Voice cloning is not available"

This message appears when voice cloning is not enabled for your account. Contact support to enable it.

"You've reached the maximum of 10 custom voices"

Delete an existing voice before creating a new one:

Go to Voice Library
Find a voice you no longer need
Click the trash icon to delete it
Try creating your new voice again

Voice sounds robotic or unnatural

Causes and fixes:

Low quality samples: Re-record with better microphone/environment
Scripted reading: Provide samples with natural conversational speech
Too short duration: Add more samples to reach 1-2 minutes total

Preview doesn't play

Check browser audio permissions
Try a different browser
Regenerate the preview sample
Ensure your device volume is on

Cloning fails repeatedly

Common causes:

Audio files are corrupted—try converting to a different format
Background noise too heavy—record cleaner samples
File format not supported—ensure MP3, WAV, or M4A
File exceeds 10MB—compress or trim the audio

Voice doesn't match original speaker

Voice cloning captures general characteristics but isn't perfect reproduction. Results depend on:

Sample quality and quantity
Uniqueness of the speaker's voice
Background noise levels
Consistency across samples

For critical brand voices, test with a few video generations before full production use.

Privacy and Data

Audio Sample Storage

Samples are uploaded to secure cloud storage
Files are private and only accessible to your account
Samples are retained while the voice exists
Deleting a voice removes associated samples

Voice Clone Ownership

Custom voices are tenant-specific (not shared across accounts)
ILLIXIS does not share voice data between customers

Consent Considerations

Before cloning someone's voice:

Ensure you have their consent for commercial use
Voice cloning for deception or impersonation violates terms of service
Use only voices you have rights to reproduce

Related Features

Voice Library - Browse and manage all voices (preset and custom) in Settings > Brand > Voice Library
AI Video Generation - Use custom voices for video narration
Video Production - Full storyboard workflow with voice selection
Brand Music - Custom audio for video end cards (separate from voiceovers)
Preset Voices — High-quality AI voices available to all users without cloning

Custom Voice Cloning