Skip to main content
Prerequisites: Have your agent created first.
Access thousands of professional voices from leading AI voice providers to give your agents unique personalities and optimize them for specific tasks and audiences.

Voice Marketplace Overview

Thoughtly provides access to an extensive library of voices from industry-leading platforms including ElevenLabs and Cartesia. Each voice offers unique characteristics in terms of gender, language, accent, and personality traits to match your specific business needs.

Voice Characteristics

The voice marketplace displays voices with the following attributes:
AttributeDescription
NameUnique identifier for each voice
GenderMale, Female, or Neutral options
LanguagePrimary language and regional accent
PlatformVoice provider (ElevenLabs or Cartesia)
ActionsPreview and Save options
Voice marketplace showing voices with characteristics and action buttons

Voice marketplace table

Browsing and Previewing Voices

Accessing the Voice Library

  1. Navigate to Voices in the sidebar
  2. Wait for the voices table to load
  3. Browse available voices using the table interface

Voice Preview

Before selecting a voice for your agents:
  1. Click the Preview button in the Actions column
  2. Listen to the voice sample (typically 10-15 seconds)
  3. Evaluate voice quality, tone, and suitability for your use case

Saving Voices for Agent Use

The Save Process

Saving vs. Default Setting: Clicking Save does NOT:
  • Make the voice the default for new agents
  • Change existing agent voice assignments
  • Apply the voice automatically to any agents
What Save Actually Does:
  • Makes the voice available in the Agent Builder voice selection dropdown
  • Adds the voice to your workspace’s saved voice library
  • Enables the voice for assignment to current and future agents
Voice availability depends on the vendor and can change without notice. Voices may be removed from the marketplace by providers at any time. Thoughtly does not control voice availability or guarantee long-term access to specific voices.

Step-by-Step Voice Saving

  1. Preview the voice to confirm it meets your needs
  2. Click Save next to the Preview button
  3. Repeat for multiple voices as needed for different agent personalities
Thoughtly Voice Library

Applying Saved Voices to Agents

Agent Voice Assignment

After saving voices, assign them to specific agents:
  1. Navigate to your agents list
  2. Open the Agent Builder by clicking the agent name
  3. Locate the voice selection in the right sidebar
  4. Choose from your saved voices in the dropdown menu
  5. Save the agent configuration
Result: The agent will use the selected voice for all future calls and interactions.

Voice Selection Strategy

Consider Agent Role: Match voice characteristics to agent function (professional for sales, friendly for support) Audience Alignment: Choose voices that resonate with your target customer demographics Brand Consistency: Ensure voice selections align with your brand personality Testing Approach: Try different voices and gather feedback on effectiveness
Agent Builder voice selection dropdown with saved voices

Agent voice assignment

Voice Vendor Considerations

FeatureElevenLabsCartesia
Voice QualityHigh-fidelity, natural-sounding voicesOptimized for real-time voice generation
LatencyStandard processing timeUltra-low latency voice processing
Voice SelectionWide selection of accents, languages, and personalitiesGrowing library of professional voices
Best ForNatural, expressive conversationsReal-time applications requiring speed
LicensingSubject to ElevenLabs terms and availabilitySubject to Cartesia terms and availability

Vendor Licensing

Users must accept any licensing constraints imposed by voice vendors. Voice usage is subject to the terms of service of the respective voice provider.

Language and Localization

Supported Languages

The voice marketplace includes voices in multiple languages to support international operations and diverse customer bases. Common languages include:
  • English (various accents: US, UK, Australian, etc.)
  • Spanish (Latin American and European variants)
  • French (French and Canadian variants)
  • German, Italian, Portuguese, and many others

Voice-Language Pairing

Important: For optimal performance, match voice language with your agent’s intended conversation language. Using voices in languages different from your agent’s conversation language may result in poor pronunciation or unnatural speech patterns.

Voice Management

Workspace-Level Voices

Saved voices are associated with your workspace, not individual users:
  • All workspace members can access saved voices
  • Voice selections persist across agent configurations
  • No sharing limitations between team members

Voice Organization

Naming Strategy: Use descriptive voice names to identify suitable use cases Documentation: Keep notes on which voices work best for specific agent types Regular Review: Periodically evaluate voice performance and customer feedback

Expected Results

After successful voice selection and assignment: Enhanced Agent Personality:
  • Agents sound more professional and aligned with your brand
  • Consistent voice quality across all customer interactions
  • Distinct agent identities for different roles or services
Improved Customer Experience:
  • More engaging and natural-sounding conversations
  • Better customer connection through appropriate voice selection
  • Professional presentation across all touchpoints
Operational Benefits:
  • Easy voice updates without changing agent logic
  • Scalable voice management across multiple agents
  • Centralized voice library for consistent branding

Limitations

Current Limitations

  • Voice availability subject to vendor changes without notice
  • No bulk voice operations (must save/assign individually)
  • No voice usage analytics or performance metrics currently available

Vendor Dependencies

  • Service availability depends on ElevenLabs and Cartesia uptime
  • Voice quality controlled by respective vendors
  • Licensing terms may change based on vendor policies

Best Practices

Voice Selection Process

Test Thoroughly: Preview multiple voices before making final selections Consider Context: Match voice tone to expected conversation types Get Feedback: Test voices with actual customers or team members when possible Plan for Scale: Select voices that work well across high call volumes

Voice Assignment Strategy

Role-Based Assignment: Use different voices for sales, support, and administrative agents Consistent Branding: Maintain voice consistency within agent categories Regular Evaluation: Monitor voice performance and customer feedback

Premium Voices

Some platforms charge a higher fee for their Voices. These Voices are marked next to the platform name with a note like “3x credits.” This means that, when you use this Voice with your Voice Agent, it will consume three times as many Credits as a standard Voice. Learn more about Credits.

Troubleshooting

Voice not appearing in Agent Builder dropdown
  • Verify you clicked “Save” (not just Preview) in the marketplace
  • Refresh the Agent Builder page
  • Check that voice is still available from vendor (see availability note)
  • Try saving a different voice to test the save functionality
Voice sounds different in production than preview
  • Preview uses sample text; actual performance varies with your prompts
  • Voice quality depends on network conditions and server load
  • Try different presence settings to optimize
  • Consider testing with Call Me before deploying
Premium voices consuming too many credits
  • Check voice multiplier (2x, 3x) shown in marketplace
  • Use premium voices only for high-value interactions
  • Switch to standard voices for high-volume use cases
  • Monitor usage in Billing dashboard
Voice pronunciation issues
Common Mistake: Not testing voices with your actual agent prompts before deploying. Preview samples may sound great, but performance varies with your specific use case. Always test with Call Me using real conversation flows.

Planned Updates

New Voice UI: A redesigned voice management interface is planned for Q1 2026, which may include enhanced browsing, filtering, and management capabilities.

See also