Voice Cloning

Create custom voice clones from your own recordings to provide highly personalized and unique agent experiences that match your brand or specific requirements.

​

Voice Cloning Overview

Thoughtly’s voice cloning feature enables you to create custom voices by recording a short audio sample. The AI system analyzes your recording and generates a voice clone that can be used across your agents, providing completely unique voice characteristics that aren’t available in the standard marketplace.

​

Voice Cloning Requirements

Minimum Recording Length: 20 seconds of clear, continuous speech Recording Method: Browser-based recording using Thoughtly’s built-in interface Real-time Processing: Recordings must be made live through the platform (no file uploads)

​

Recording Process

​

Step 1: Access Voice Cloning

1. Navigate to Voices in the sidebar
2. Click “Clone Voice” button on the voices page
3. Review the cloning modal and requirements

​

Step 2: Prepare for Recording

Environment Setup:

• Quiet space: Use a noise-free environment for best results
• Quality microphone: Built-in laptop mics work, but external mics provide better quality
• Consistent tone: Maintain steady volume and speaking pace
• Clear articulation: Speak clearly and avoid mumbling

Content Preparation:

• Script planning: Prepare 20+ seconds of natural, conversational content
• Varied speech: Include different sentence structures and tones
• Brand alignment: Use language and tone consistent with your desired agent personality

​

Step 3: Record Your Voice Sample

1. Click the record button in the cloning modal
2. Speak naturally for at least 20 seconds
3. Include variety in tone, pace, and sentence structure
4. Stop recording when you have sufficient content
5. Review the recording quality before proceeding

​

Step 4: Process and Name

1. Enter a descriptive name for your custom voice
2. Submit the recording for AI analysis
3. Wait for processing (typically several minutes)

Processing Time: Voice clone generation may take longer than standard content processing due to the complexity of voice synthesis model creation.

​

Recording Best Practices

​

Audio Quality Guidelines

Clear Speech: Articulate words clearly and avoid rushed speech Consistent Volume: Maintain steady speaking level throughout the recording Natural Intonation: Use varied sentence patterns and natural conversation flow Avoid Background Noise: Record in quiet environments without echo or interference

​

Content Guidelines

Conversational Style: Record in the tone and style you want agents to use Complete Sentences: Include full thoughts and natural speech patterns Emotional Range: Demonstrate slight variations in tone appropriate for customer service Professional Tone: Maintain appropriate business communication style

​

Technical Considerations

Browser Compatibility: Ensure your browser supports microphone access and audio recording Microphone Permissions: Grant necessary permissions when prompted Stable Connection: Maintain reliable internet during recording and processing

​

Voice Clone Management

​

Workspace-Level Access

Custom voice clones are created per workspace:

• Private to workspace: Cloned voices are not shared with other Thoughtly users
• Team access: All workspace members can use cloned voices for their agents
• Persistent storage: Voice clones remain available until manually deleted

​

Voice Clone Availability

After successful processing:

1. Appears in the voices table alongside marketplace voices
2. Available for preview like standard voices
3. Selectable in Agent Builder voice dropdown
4. Assignable to multiple agents as needed

​

Usage and Application

​

Agent Assignment

Apply cloned voices to agents using the same process as marketplace voices:

1. Navigate to Agent Builder for your target agent
2. Open voice selection in the right sidebar
3. Choose your custom voice from the dropdown
4. Save agent configuration

​

Performance Characteristics

Quality Expectations: Voice clone quality depends on recording quality and AI processing Consistency: Cloned voices maintain characteristics across different text inputs Limitations: May not perfectly replicate all nuances of the original speaker

​

Multi-Agent Usage

Unlimited Assignment: Custom voices can be used by multiple agents simultaneously No Usage Limits: No restrictions on call volume or concurrent usage Consistent Experience: All agents using the same clone will sound identical

​

Legal and Ethical Considerations

​

Consent and Authorization

Recording Consent: Only record voices from individuals who have explicitly consented to voice cloning Usage Rights: Ensure you have proper authorization to use the recorded voice for business purposes Brand Representation: Confirm that voice characteristics align with your brand and legal requirements

​

Recommended Practices

Clear Agreements: Document consent and usage permissions for recorded voices Professional Use: Limit usage to appropriate business contexts Quality Control: Monitor agent performance to ensure voice clone represents your brand appropriately

​

Data Storage and Privacy

​

Recording Storage

Processing Only: Raw audio recordings are used for voice clone generation Workspace Isolation: Voice clones are not accessible to other Thoughtly workspaces Retention Policy: Check current data retention policies for voice clone storage duration

​

Expected Results

After successful voice cloning and assignment: Unique Brand Voice:

• Agents use completely custom voice not available to other businesses
• Consistent brand representation across all agent interactions
• Professional, personalized customer experience

Operational Benefits:

• Distinctive agent personalities that align with brand identity
• Memorable customer interactions through unique voice characteristics
• Scalable custom voice deployment across multiple agents

​

Limitations

​

Technical Limitations

Recording Requirements: Must use browser-based recording (no file upload option) Processing Time: Voice clone generation takes longer than standard operations Quality Variables: Final voice quality depends on recording conditions and source material

​

Usage Constraints

Single Recording: Each voice clone based on one 20-second recording session No Editing: Cannot modify or refine existing voice clones Re-recording Required: Changes require creating entirely new voice clone

​

Troubleshooting

Recording fails to start

• Check browser microphone permissions
• Verify microphone functionality in browser settings
• Try different browser if issues persist

Poor voice clone quality

• Re-record with better audio quality
• Ensure 20+ seconds of clear, varied speech
• Check for background noise or audio interference

Processing takes too long

• Wait for completion (processing can take several minutes)
• Check system status if processing exceeds expected timeframes
• Contact support with Team ID if processing fails

Voice clone not appearing in Agent Builder

• Verify voice clone processing completed successfully
• Refresh Agent Builder interface
• Confirm voice clone appears in main voices table