Voice Setup
Enable your AI persona to speak with your voice, creating an even more authentic connection with your audience.
Voice Features Overview
| Feature | Description |
|---|---|
| Voice Cloning | AI learns your unique voice |
| Audio Responses | Persona speaks responses aloud |
| Voice Conversations | Real-time voice chat |
| Voice Messages | Async audio replies |
Voice Cloning
How It Works
Your persona learns your voice from:
- Video audio tracks
- Podcast episodes
- Voice recordings
- Any audio content you provide
The AI captures:
- Your tone and pitch
- Speech patterns
- Pronunciation
- Pacing and rhythm
- Emotional expression
Minimum 30 Minutes of Audio Required
Voice cloning quality depends heavily on the amount and variety of audio you provide. Uploading less than 30 minutes of audio will produce a noticeably lower-quality voice clone that may sound robotic or unlike you.
Audio Requirements
| Requirement | Specification |
|---|---|
| Total Duration | Minimum 30 minutes recommended |
| Quality | Clear audio, minimal background noise |
| Variety | Different topics and emotions |
| Format | MP3, WAV, M4A, or video with audio |
Recording Best Practices
For Best Voice Quality:
- Use a quality microphone
- Record in a quiet space
- Speak naturally (don't read robotically)
- Include emotional variety
- Cover your typical topics
Voice Training Process
-
Upload Audio
- Go to Voice > Voice Training
- Upload audio files or connect content
- Processing begins automatically
-
Review Samples
- Listen to AI-generated samples
- Compare to your actual voice
- Note any issues
-
Fine-Tune
- Adjust voice settings
- Add more audio if needed
- Iterate until satisfied
-
Approve
- Confirm voice quality
- Enable for your persona
- Set availability options
Voice Quality Scores
| Score | Quality | Action |
|---|---|---|
| 90-100 | Excellent | Ready to use |
| 75-89 | Good | Minor differences |
| 60-74 | Fair | Consider more training |
| Below 60 | Needs work | Add more/better audio |
Voice Response Settings
When to Use Voice
| Setting | Description |
|---|---|
| Always | Every response includes audio |
| On Request | Only when fan asks |
| Auto-Detect | Based on conversation context |
| Never | Text only |
Response Modes
| Mode | How It Works |
|---|---|
| Text + Audio | Both text and voice provided |
| Audio Only | Voice response, no text shown |
| Text First | Text immediate, audio follows |
| Audio First | Voice plays, text appears |
Audio Length Limits
| Setting | Max Duration | Best For |
|---|---|---|
| Brief | 30 seconds | Quick answers |
| Standard | 2 minutes | Most responses |
| Extended | 5 minutes | Detailed explanations |
| Unlimited | No limit | Special content |
Voice Conversation Mode
Real-Time Voice Chat
Enable fans to have voice conversations:
How It Works:
- Fan initiates voice chat
- They speak their message
- Persona responds with voice
- Natural back-and-forth conversation
Voice Chat Settings
| Setting | Options |
|---|---|
| Availability | Always, Scheduled, Subscriber Only |
| Turn Detection | Automatic, Manual |
| Response Delay | Instant, Natural pause |
| Interruption Handling | Allow, Queue, Disable |
Voice Chat Quality
Optimize for best experience:
| Factor | Recommendation |
|---|---|
| Latency | Aim for under 1 second |
| Audio Quality | High bitrate when possible |
| Fallback | Text backup if audio fails |
Voice Messages
Asynchronous Voice
Fans can send voice messages:
- Fan records voice message
- AI transcribes and understands
- Persona responds with voice
- Fan listens when convenient
Voice Message Settings
| Setting | Description |
|---|---|
| Max Length | How long fans can record |
| Auto-Play | Response plays automatically |
| Transcription | Show text of fan's message |
Voice Customization
Voice Parameters
Fine-tune your cloned voice:
| Parameter | Range | Effect |
|---|---|---|
| Speed | 0.5x - 2.0x | Speaking pace |
| Pitch | -10 to +10 | Voice pitch adjustment |
| Expressiveness | Low to High | Emotional range |
| Clarity | Standard to Enhanced | Articulation |
Emotion Settings
Configure emotional expression:
| Emotion | When Used |
|---|---|
| Neutral | Standard responses |
| Excited | Positive topics, enthusiasm |
| Thoughtful | Complex questions |
| Empathetic | Supportive moments |
| Playful | Humorous content |
Pronunciation Guides
Add custom pronunciations:
- Go to Voice > Pronunciation
- Click Add Word
- Enter word and phonetic guide
- Test in preview
- Save
Common Uses:
- Your name
- Brand names
- Technical terms
- Inside jokes
Multilingual Voice
Multiple Languages
If you speak multiple languages:
- Upload audio in each language
- Train voice separately
- Set language detection
- Persona switches automatically
Language Settings
| Setting | Description |
|---|---|
| Primary | Default language |
| Secondary | Additional languages |
| Detection | Auto-detect fan's language |
| Fallback | Language if detection fails |
Voice Analytics
Voice Usage Metrics
| Metric | Description |
|---|---|
| Voice Requests | How often voice is used |
| Completion Rate | Audio listened to completion |
| Preference | Voice vs. text preference |
| Quality Feedback | Fan ratings |
Viewing Analytics
- Go to Analytics > Voice
- See usage trends
- Review quality metrics
- Identify improvement areas
Voice Privacy and Safety
Content Filtering
Voice responses filtered for:
- Inappropriate content
- Personal information
- Boundary violations
Voice Watermarking
Optional invisible watermark:
- Identifies AI-generated audio
- Helps prevent misuse
- Transparent to listeners
AI Transparency Required
Fans must always be informed that they are interacting with an AI-generated voice, not the real creator. This is both a platform requirement and a legal best practice in many jurisdictions.
Usage Consent
Configure consent requirements:
| Setting | Description |
|---|---|
| Notice | Inform fans it's AI voice |
| Opt-In | Fans must enable voice |
| Terms | Link to voice usage policy |
Troubleshooting
Common Issues
| Issue | Cause | Solution |
|---|---|---|
| Voice sounds robotic | Not enough training | Add more diverse audio |
| Wrong pronunciation | Missing pronunciation guide | Add custom pronunciations |
| Audio cuts off | Length limits | Adjust response length |
| Quality varies | Inconsistent training | Use consistent audio quality |
Voice Quality Tips
- Upload at least 30 minutes of audio
- Use varied content (different emotions, topics)
- Ensure clean audio without background noise
- Include natural speech, not just reading
- Update periodically with new content
Best Practices
Creating Natural Voice
- Train with conversational audio
- Include emotional variety
- Use authentic content
- Test with real questions
Fan Experience
- Set appropriate expectations
- Provide text alternative
- Monitor feedback
- Iterate based on usage
Next Steps
- Publish Your Persona - Launch with voice
- Fan Engagement - Engage with voice
- Monetization - Premium voice features