Inline Voice Preview in Assistant Configuration
Overview
The Inline Voice Preview feature enables designers and administrators to listen to voice samples directly from the Assistant Configuration screen without creating or executing a flow. This simplifies voice selection by providing instant audio playback within the configuration experience.
Previously, users had to build and run a complete flow through the Runtime Engine to hear a voice. With Inline Voice Preview, voices can be evaluated in seconds, making voice selection faster, easier, and more efficient.
Key Benefits
Faster Assistant Configuration: Eliminates the need to create and test a flow simply to hear a voice, reducing setup time significantly.
Improved Voice Selection: Allows designers to make informed decisions by listening to voice samples before assigning them to assistants.
Faster Provider Comparison: Enables quick comparison of voices across supported providers, helping users select the most suitable voice.
Better User Experience: Keeps voice selection and validation within the Assistant Configuration page, avoiding unnecessary navigation and context switching.
Self-Contained Configuration
Voice preview remains part of the assistant setup process, reducing dependency on Runtime Engine test sessions.
Prerequisites
Before using Inline Voice Preview:
Access to the Admin Dashboard.
Permission to configure assistants.
A supported text-to-speech provider configured in the environment.
An active internet connection for audio playback.
Supported Voice Providers
Inline Voice Preview supports all voices available through the platform voice catalog, including:
Azure Text-to-Speech
Platform-supported Azure TTS voices, including Azure OpenAI TTS models.
ElevenLabs
Standard and custom voices available through Eleven Labs integration.
Future Providers
Additional TTS providers may be supported as they are onboarded into the platform.
How Voice Preview Works
To maintain security and protect provider credentials:
The user selects a voice from the Assistant Configuration screen.
The user enters a short sample sentence.
The user clicks Play.
The Admin Dashboard sends the selected voice information and sample text to the Configuration Service.
The Configuration Service securely communicates with the configured TTS provider.
Audio is synthesized server-side.
The generated audio is streamed back to the browser.
The browser plays the audio immediately without downloading any files.
All requests are secured using the platform's authentication and tenancy controls.
Accessing Voice Preview
Open the Admin Dashboard.
Navigate to the desired assistant.
Open the Assistant Configuration page.
Locate the Voice Selection section.
Select a voice from the available catalog.
The Sample Text field and Play option become available within the voice selection area.
Previewing a Voice
Step 1: Select a Voice
Choose a voice from the catalog based on:
Language
Vendor
Accent
Voice style
Step 3: Play the Preview
Select Play.
The selected provider synthesizes the text and plays the generated audio directly in the browser.
Step 4: Compare Voices
Change the voice or provider and replay the audio as needed until you find the preferred voice.
Step 5: Save Your Selection
Once satisfied with the voice:
Select the preferred voice.
Save the assistant configuration.
Publish the assistant or associated flow.
Error Handling
If audio generation or playback fails, the system displays a clear error message without disrupting the configuration process.
Common scenarios include:
Voice Generation Failure: The selected provider is temporarily unavailable.
User Action:
Retry the preview.
Select a different voice.
Contact an administrator if the issue persists.
Network Interruption: The browser is unable to receive the generated audio stream.
User Action:
Check network connectivity.
Retry playback.
Invalid Sample Text: The entered text exceeds permitted limits or contains unsupported content.
User Action:
Shorten or modify the text.
Retry playback.
Example Use Cases
Customer Service Assistants: Select a professional and approachable voice for customer-facing interactions.
Virtual Receptionists: Evaluate friendly and welcoming voice options.
Sales Assistants: Choose a voice that reflects the organization's brand personality.
Multilingual Assistants: Compare provider voices across different languages and regional accents.
Custom Voice Deployments: Validate generated or imported voices before assigning them to production assistants.
Security and Compliance
Voice preview requests are processed securely through the Configuration Service.
Key security measures include:
Provider credentials remain server-side.
No credentials are exposed to the browser.
Standard platform authentication is enforced.
Organization-level tenant isolation is maintained.
Preview operations follow existing platform security and audit controls.
Summary
The Inline Voice Preview feature streamlines voice selection by allowing users to listen to synthesized speech directly from the Assistant Configuration screen. By eliminating the need to build and execute test flows, designers can quickly compare voices, evaluate audio quality, and confidently select the most appropriate voice for their assistants, resulting in a faster and more efficient configuration experience.
Last updated
Was this helpful?