---
"@context": https://schema.org
"@type": TechArticle
"@id": https://www.twilio.com/docs/voice/voice-insights/api/call/details-conversation-relay-events#article
headline: "Details: Conversation Relay Insights Events"
description: Supplementary details and explanations of Conversation Relay events in Voice Insights. Reference for event types, latency signals, and interaction markers available via the Call Insights Events API for calls using Conversation Relay.
url: https://www.twilio.com/docs/voice/voice-insights/api/call/details-conversation-relay-events
inLanguage: en
dateModified: 2026-07-07T17:14:48.000Z
author:
  "@type": Organization
  name: Twilio Developer Education Team
publisher:
  "@type": Organization
  name: Twilio
---

# Details: Conversation Relay Insights Events

This page is a reference for all event types emitted by Conversation Relay and available via the [Call Insights Events resource](/docs/voice/voice-insights/api/call/call-events-resource).

> \[!NOTE]
>
> Conversation Relay events are emitted at the `carrier_edge`. If your call type doesn't default to `carrier_edge`, pass `Edge=carrier_edge` when requesting events. Otherwise, Conversation Relay events aren't returned.

All Conversation Relay events include the following base properties:

| **Property**      | **Type** | **Description**                               |
| ----------------- | -------- | --------------------------------------------- |
| `session_id`      | string   | Session ID of the Conversation Relay session. |
| `sequence_number` | integer  | Sequence number of the event.                 |

Events with additional payload properties are documented below. Events not listed in the payload sections are timeline markers only and carry no additional properties beyond the base fields.

## Speech events

| **Event**                  | **Description**                                                            |
| -------------------------- | -------------------------------------------------------------------------- |
| `start_of_customer_speech` | Customer audio detected — beginning of an utterance.                       |
| `end_of_customer_speech`   | Customer stopped speaking; transcription will be sent to your application. |
| `start_of_agent_speech`    | Virtual agent began playing audio to the customer.                         |
| `end_of_agent_speech`      | Virtual agent finished its current response.                               |

## Token events

| **Event**              | **Description**                                                                                |
| ---------------------- | ---------------------------------------------------------------------------------------------- |
| `first_token_received` | First token received from the application via WebSocket. Marks the end of application latency. |
| `final_token_received` | Full text response received from the application.                                              |

### final\_token\_received properties

| **Property**   | **Type** | **Description**                  |
| -------------- | -------- | -------------------------------- |
| `total_tokens` | integer  | Total number of tokens received. |
| `total_words`  | integer  | Total number of words received.  |

## Latency events

| **Event**     | **Description**                                                  |
| ------------- | ---------------------------------------------------------------- |
| `stt_latency` | Time from end of customer utterance to full transcription ready. |
| `tts_latency` | Time from receiving a text sentence to start of audio synthesis. |

Both latency events share the same payload:

| **Property** | **Type** | **Description**          |
| ------------ | -------- | ------------------------ |
| `latency_ms` | integer  | Latency in milliseconds. |

## Interaction events

| **Event**     | **Description**                                                                |
| ------------- | ------------------------------------------------------------------------------ |
| `prompt_sent` | Transcription sent to your WebSocket server.                                   |
| `interrupt`   | One party began speaking while the other was active.                           |
| `preempted`   | New agent response tokens arrived, interrupting current agent speech playback. |
| `dtmf`        | Customer pressed a key on their phone's keypad.                                |
| `digit`       | Application sent digits to navigate an IVR.                                    |

### interrupt properties

| **Property** | **Type** | **Description**                                  |
| ------------ | -------- | ------------------------------------------------ |
| `type`       | string   | Type of interruption. One of `DTMF` or `SPEECH`. |

## Session events

| **Event**          | **Description**                                           |
| ------------------ | --------------------------------------------------------- |
| `configurations`   | Session configuration active at the start of the session. |
| `language_changed` | Active language configuration changed during the session. |
| `call_wrap_up`     | Session has ended and the summary is being finalized.     |

### configurations properties

| **Property**     | **Type** | **Description**                                                                                                                    |
| ---------------- | -------- | ---------------------------------------------------------------------------------------------------------------------------------- |
| `configurations` | object   | Key-value pairs of active configuration settings.                                                                                  |
| `languages`      | object   | Key-value pairs of language configurations, each containing `tts_provider`, `voice`, `transcription_provider`, and `speech_model`. |

### language\_changed properties

| **Property**                  | **Type** | **Description**                   |
| ----------------------------- | -------- | --------------------------------- |
| `tts_language_code`           | string   | Language code for text-to-speech. |
| `transcription_language_code` | string   | Language code for speech-to-text. |

### call\_wrap\_up properties

| **Property**          | **Type** | **Description**                                                                 |
| --------------------- | -------- | ------------------------------------------------------------------------------- |
| `duration_in_seconds` | integer  | Session duration in seconds.                                                    |
| `end_status`          | string   | Final state of the session. One of `unknown`, `failure`, `ended`, or `hung_up`. |

## Media events

| **Event**    | **Description**                                                                 |
| ------------ | ------------------------------------------------------------------------------- |
| `play_media` | Application requested an audio file to be played instead of synthesized speech. |

## Error events

| **Event**     | **Description**                                   |
| ------------- | ------------------------------------------------- |
| `error_event` | Technical failure encountered during the session. |

### error\_event properties

| **Property** | **Type** | **Description** |
| ------------ | -------- | --------------- |
| `error_code` | integer  | Error code.     |
| `message`    | string   | Error message.  |