Table of content

Analytics events

This feature is only available as part of a closed beta for the time being.

V1 protocol users should refer to the documentation for the legacy payloads.

During the conversation, various NLP processors are run according to your subscription. They emit annotation events that interpret and enrich the transcript. Those annotations are anchored on the transcript timeline, so that:

  • they can also apply to non-speech sound,
  • you can rewrite the transcript according to the interpretation in order to improve readability, or to hide sensitive information.

Events described below are Analytics events. Such events are returned when the text or the audio signal have matched an analytics rule.

Tags

The result of any analytics is named a tag in our API. Events sent to the client are therefore named tags_set. Note that several tags can be returned at once.

The payload is:

 {
"campaign_id": "tests",
"client_id": "acme",
"components":
[
{
"confidence": 0.8,
"display": "Merci",
"value": "fedcba90-5175-4c8b-8378-d7237c5271c2"
},
{
"confidence": 0.8,
"display": "Speaker Alice",
"value": "aaaaaaaa-5175-4c8b-abcd-d7237c5271c2"
}
],
"conversation": "conversation:acme_corp@conference",
"country": "fr",
"end": 1761930327034,
"id": "tags-1761930326254",
"lang": "fr",
"length": 780,
"speaker": "Alice",
"start": 1761930326254
}