Provide a live audio streaming source, can be a video or an audio bitstream.

To request a tagging session, call the API path below. This process is automatic, and you do not have to perform any additional steps to start the session.


HEADER x-chatsight-api-auth: <API-TOKEN>
HEADER x-chatsight-api: chatsightvoice

curl --location --request POST '' \
--header 'x-chatsight-api-auth: <YOUR_API_KEY>' \
--header 'x-chatsight-api: chatsightvoice' \
--header 'Content-Type: application/json' \
--data-raw '{
urlSTRINGThe pre-signed, or publicly available URL to a media stream. If the source is a video, we encourage you to provide a URL to a low-quality version (360p and below), to reduce turn around times and avoid OOM issues.
convoIdSTRINGThe Conversation Identifier for this session. We encourage you to use a collision-safe naming system, like UUID. Using standard words without unique elements will OVERWRITE the existing entry.
webhookSTRINGExplain below, provide a valid HTTP POST endpoint to receive results by webhook additionally.
seedSentenceSTRINGA UTF-8 string of a textual seed sentence to improve tagging recommendations.


Instead of polling the query endpoint described below, we offer a "one-shot" webhook feature. This means that upon the completion of the session, a single webhook POST request will be made against the target URL of your choosing.

Your target URL must return a response that is exactly a HTTP 200 OK response.

The webhook payload that would be sent to the endpoint is structured like this:

  "sender": "chatsight-webhooks",
  "addon": "ssiQuery",
  "payload": {
    "tags": "Fun,Relaxed", // The resulting tags (explained in "Query a Tagging Session").
    "undocumentedTags": false, // Tags returned, that are not documented, (explained in "Query a Tagging Session").
    "convoId": "pizza_stream" // Conversation Identifier.
  "time": 1620487495836, // Time of the request, in MS.
  "hmac": "ZdEWroII9eiNmfb0qtp2lcMhsvhjBaoejR7lwiVQcoQ=" // HMAC.