Skip to Content
Node ReferenceTTS Response

TTS Response

Convert text to spoken responses - deliver custom messages to callers at any point in your flow.

Overview

The TTS Response node speaks a custom message to callers using HD voice synthesis. Use it to deliver confirmations, provide information, or guide callers through your flow.

Unlike the Start node (which greets callers), TTS Response nodes can be placed anywhere in your flow to speak additional messages during the conversation.

Configuration

Response Text

Type: string Required: Yes

The message that will be spoken to the caller. You can include dynamic variables to personalize the response based on conversation context.

Example:

Thank you for your question about {{user_input}}. Let me help you with that.

Available Variables

When expanded, you can click to insert dynamic variables that get replaced with actual values during the call:

VariableDescription
{{user_input}}What the caller just said
{{confidence}}How confident the AI is in its understanding (0-100%)
{{reasoning}}The AI’s reasoning for its response
{{matched_question}}The closest matching FAQ question (if using FAQ Loader)
{{faq_answer}}The answer from the matched FAQ

Example with variables:

I understand you're asking about {{matched_question}}. Here's what I found: {{faq_answer}}

How It Works

  1. The call reaches the TTS Response node from the previous step
  2. Any variables in your response text are replaced with actual values
  3. The message is converted to speech using HD voice synthesis
  4. The spoken message plays to the caller
  5. The call automatically continues to the next connected node

Connection Rules

DirectionAllowedNotes
Inputs1Receives connection from one previous node
Outputs1Connects to one next node in the flow

TTS Response is a simple pass-through node - it speaks the message and immediately continues to the next step. No caller input is required.

Examples

Basic: Simple Follow-up Message

A confirmation message after processing a question:

Configuration:

  • Response Text: Got it. Let me look that up for you.

Flow pattern:

Query Processor --> TTS Response --> [Next Step]

Advanced: Dynamic Response with Variables

A personalized response using conversation context:

Configuration:

  • Response Text: You asked about {{matched_question}}. {{faq_answer}}

Flow pattern:

FAQ Loader --> Query Processor --> TTS Response --> End

Transfer Announcement

Let callers know they’re being transferred:

Configuration:

  • Response Text: I'll connect you with a specialist who can help. Please hold while I transfer you.

Flow pattern:

Query Processor --> TTS Response --> Transfer

Goodbye Message

A farewell before ending the call:

Configuration:

  • Response Text: Thank you for calling. Have a great day!

Flow pattern:

Query Processor --> TTS Response --> End

Best Practices

  • Keep messages concise - Aim for 2-3 sentences (under 200 characters)
  • Use variables for personalization - Make responses feel natural by referencing what the caller said
  • Configure voice in Agent Settings - Voice selection is done at the agent level, not per-node
  • Chain multiple TTS nodes if needed - For longer content, use multiple nodes for better flow control
  • Test with the Chat feature - Preview how your messages sound before going live

Good Response Text

"I found the answer to your question. {{faq_answer}}"
"Please hold while I transfer you to our sales team."

Avoid

  • Very long messages (split into multiple nodes)
  • Technical jargon or internal terms
  • Repeating information the caller already knows
  • Empty or placeholder text

Common Issues

Variables not being replaced

Problem: You see {{user_input}} spoken literally instead of the actual value.

Solution:

  1. Make sure the variable name is spelled correctly (case-sensitive)
  2. Verify the previous node (like Query Processor) is passing the variable data
  3. Check that variable syntax uses double curly braces: {{variable_name}}

Response is too long

Problem: The message gets cut off or sounds rushed.

Solution: Break the message into multiple TTS Response nodes. This gives callers time to process each piece of information.

Voice sounds different than expected

Problem: The TTS voice doesn’t match what you configured.

Solution: Voice settings are configured in Agent Settings, not on individual TTS nodes. Go to the Agent tab in the sidebar to update voice preferences.

Message not playing

Problem: The node is reached but no message plays.

Solution:

  1. Verify the Response Text field is not empty
  2. Check the flow is properly connected
  3. Ensure the agent is synced (save and republish if needed)
  • Start - The initial greeting node; TTS Response handles follow-up messages
  • Query Processor - Provides variables like {{user_input}} and {{confidence}}
  • End - Often preceded by a TTS Response for a goodbye message
  • Transfer - Commonly paired with TTS Response for transfer announcements
Last updated on