Metadata Type: AiPlannerVoiceDef
AiPlannerVoiceDef is a metadata type in Salesforce that represents the voice definition for an Agentforce voice agent. This metadata type allows organizations to configure and store the speech-to-text and text-to-speech settings used during AI voice interactions, giving administrators control over how their automated agents sound and interpret spoken language.
Overview
The AiPlannerVoiceDef metadata type is part of Salesforce's efforts to enhance AI-driven voice capabilities within the Agentforce ecosystem. It enables administrators to define exact parameters for audio synthesis, select specific AI models for transcription, and assign distinct voice personas to their agents. This functionality is particularly useful for organizations looking to implement intelligent, conversational voice assistants for customer service, sales, and internal automation.
Key Features
- Configuration for speech-to-text and text-to-speech operations
- Customizable voice personas (e.g., tone, accent, and gender)
- Audio synthesis parameter controls
- Seamless integration with Agentforce voice agents
Structure and Components
An AiPlannerVoiceDef typically consists of the following components:
- Model Selection: The specific underlying AI models used for transcribing user audio and generating agent speech
- Voice Persona: The predefined characteristics of the synthetic voice used by the agent
- Audio Synthesis Parameters: Technical settings that dictate attributes such as speech rate, pitch, and audio output format
- Language Settings: The defined language and regional locales supported by the voice definition
Deployment Considerations
When deploying AiPlannerVoiceDef metadata, administrators should be aware of several potential issues:
1. Feature Enablement
Ensure that Agentforce and its associated voice capabilities are fully enabled in the target org. If these prerequisite features are not turned on, the metadata type may not be recognized, resulting in deployment failures.
2. Model and Provider Availability
Verify that the specific AI models referenced for speech-to-text and text-to-speech exist and are licensed in the destination environment. Missing models or unconfigured providers will cause runtime errors during voice interactions.
3. Language Support
Check that the languages and locales specified in your voice definition are supported by the target environment's chosen models. Unsupported language codes can lead to failed audio synthesis or inaccurate transcription.
4. Existing Voice Agents
Be cautious when modifying existing voice definitions, as changes to the persona or speech rate will immediately affect live Agentforce agents. Consider creating new versions or thoroughly testing in a sandbox environment before deploying to production.
5. Component Dependencies
Ensure that the AiPlannerVoiceDef is properly linked to its corresponding Agentforce agent metadata. Deploying a voice definition without the underlying agent structure may lead to orphaned configurations.
Best Practices for Salesforce Administrators
To effectively use and manage AiPlannerVoiceDef metadata, Salesforce administrators should follow these best practices:
1. Document Your Configurations
Maintain clear documentation of all voice definitions, including the selected models, persona characteristics, and the specific agents they are assigned to. This is crucial for maintaining a consistent customer experience.
2. Use Descriptive Names
Choose clear and descriptive names for your AiPlannerVoiceDef instances (e.g., "Support_Agent_US_English_Female"). This makes it easier to understand their purpose at a glance and simplifies management across multiple agents.
3. Align with Brand Identity
Work closely with your organization's marketing and customer experience teams to ensure the selected voice persona accurately reflects your brand's tone, professionalism, and identity.
4. Leverage Sandboxes
Always test new or modified voice configurations in a sandbox environment. Call into the sandbox agent to verify that the speech synthesis sounds natural and the transcription accurately captures diverse accents before deploying to production.
5. Monitor and Optimize
Regularly review transcripts and audio logs of Agentforce voice interactions. Look for opportunities to tweak audio synthesis parameters, such as adjusting the speech rate if users frequently ask the agent to repeat itself.
6. Version Control
Use a version control system to track changes to your AiPlannerVoiceDef metadata. This practice allows you to quickly roll back to a previous voice configuration if a newly deployed model underperforms.
7. Implement Gradually
If you're introducing new voice agents to your org, route a small percentage of calls to the AI agent initially. Expand the routing as you verify the accuracy of the transcription and the quality of the synthesized voice.
8. Regular Audits
Conduct periodic audits of your voice definitions to ensure they remain relevant. Remove any experimental or unused voice configurations to keep your metadata footprint clean.
9. Fallback Handling
Ensure your Agentforce configuration includes robust error handling to seamlessly route users to a human representative if the voice transcription fails or the text-to-speech model times out.
10. Training and Documentation
Provide training and comprehensive documentation for the teams responsible for building conversational AI flows. Ensure they understand how to write prompt responses that sound natural when processed by the text-to-speech engine.
Conclusion
The AiPlannerVoiceDef metadata type is a powerful tool for Salesforce administrators to customize the auditory experience of their Agentforce voice implementations. By understanding its capabilities, potential deployment issues, and following best practices, administrators can effectively design natural, on-brand voice interactions for their users. As with any AI configuration, careful persona selection, rigorous auditory testing, and ongoing optimization are key to successful implementation and maintenance of AiPlannerVoiceDef metadata.