Reports the TTS Playback URLs to VA for each message received.
Tracks whether each reply was actually played out to the customer or interrupted by a barge-in.
Manages several runtime parameters that allow-disable barge-in and silence checks.
Parameters
name |
description | default |
|---|---|---|
| Address | Address of the VA Orchestrator Service. | http://cai-magpie-api.sestek-cai/magpie/ext-api/messages/ |
| Credential | OAuth 2.0 credential used to authenticate with the VA Orchestrator Service. | [Default] |
| Ignore SSL Errors | Ignore any certificate errors if Address contains "https". This is useful for testing purposes. | false |
Inputs
Audio
none
Events
| name | description | possible nodes this event may be received from |
|---|---|---|
| VA External | Registers a new VA message to track: its TTS text (if any) and any audio/minio prerecorded-audio attachments. |
VA Sync |
| Interruptive Barge In | Marks every still-pending audio item of every tracked VA message as interrupted. | Barge In Announcer |
| SR Milestone | Marks the currently tracked VA messages with 'CustomerSpokeAfterRecord', and turns off silence checks. | SR Http, SR Websocket |
| VA Silence Detected | Turns off silence checks. | VA Silence Controller |
| Merge Audio Started / Merge Audio Ended | Resolves a tracked audio item as delivered or interrupted, correlated via Data.Context.VaId. |
Batch to Stream Audio Merger |
| AudioLogReport | Used to send the audio playback link to VA. | Audio Logger |
Outputs
Audio
none
Events
| name | description |
|---|---|
| VA TTS Report Result | Sent after the VA Orchestrator Service acknowledges the report for a message. |
| VA Audio Processing Complete | Sent once an endOfConversation has been received from VA and every tracked message's audio items have all been resolved (delivered or interrupted). Carries Data.EndOfConversation = true. |
Remarks
Runtime Parameters Managed
This node automatically updates the following session/runtime parameters as VA messages are delivered or interrupted, so other nodes can react without extra wiring:
| parameter | when it changes |
|---|---|
| VaCurrentProcessingMessageId | Set to the id of the VA message currently being played out, or null when none is pending. |
| BypassFillerWordFilter | true while no VA message is pending, false while one is being played out. |
| VaVoiceInputEnabled, VaDigitInputEnabled | Set to true once a VA message has been fully delivered or fully interrupted, re-enabling customer input. |
| VaSilenceCheckEnabled | Turned on once a VA message is fully delivered or interrupted; turned off as soon as SR Milestone, VA Silence Detected or VA External activity is observed. |
Failure Safety
If the node receives an incident from TTS Http, VA Sync or Batch to Stream Audio Merger after an endOfConversation was already received, it force-emits "VA Audio Processing Complete" instead of waiting indefinitely for events that may never arrive due to the failure.
Project Structure
Needs to be connected downstream to all nodes mentioned in the Input/Events header.
Output must be able to reach the Publisher node for VA Audio Processing Complete output.
Supported flow types
Batch, Stream