Call Analysis

Retrieves the AI-generated summary and speaker-separated transcript for a specific call. The response includes the transcription and summary status, when the transcription record was created, the full summary, and each utterance in the conversation.

GEThttps://admin.zvmstelecom.net/api/v1/domains/{DOMAIN_UUID}/cdrs/{XML_CDR_UUID}/analysis

🔑

Permissions. Authenticate as described in Authentication. Your API user must have access to the requested domain and the xml_cdr_view permission.

Path Parameters

DOMAIN_UUIDstring (UUID)Required

Your domain's UUID, provided with your API credentials.

XML_CDR_UUIDstring (UUID)Required

The ID of the call you want analysis for, taken from the xml_cdr_uuid field of a List Call History result.

This endpoint takes no query parameters or request body.

Example Request

cURL

XML_CDR_UUID="YOUR_XML_CDR_UUID"

curl -s "https://admin.zvmstelecom.net/api/v1/domains/${DOMAIN_UUID}/cdrs/${XML_CDR_UUID}/analysis" \
  -H "Authorization: Bearer ${TOKEN}" \
  -H "Accept: application/json" |
  python3 -m json.tool

Response

A 200 response returns the newest transcription record for the call. If a call has more than one transcription record, you get the most recently created one, even if it is still processing.

Field	Type	Description
`xml_cdr_uuid`	string (UUID)	Unique ID of the call.
`status`	string	Transcription processing status.
`summary_status`	string or null	Summary processing status.
`created_at`	string or null	When the transcription record was created. This is not the call's start time.
`summary_payload`	object or null	The full stored summary, including any analysis fields produced by the configured summarization process.
`utterances`	array or null	Speaker-separated transcript segments. Other transcription result fields are left out.

created_at is returned as stored, typically in YYYY-MM-DD HH:MM:SS format with no time zone offset. Confirm the time zone with us before converting it.

The structure of summary_payload depends on the configured analysis format. Build your integration to tolerate added or missing summary properties.

Utterance fields

In the current transcription format, each utterance contains:

Field	Type	Description
`start`	number	Segment start time, in milliseconds from the beginning of the recording.
`end`	number	Segment end time, in milliseconds from the beginning of the recording.
`text`	string	Transcribed speech for the segment.
`speaker`	string	Speaker label within the recording, such as `A`, `B`, or `C`.
`confidence`	number	Transcription confidence reported by the provider, generally between `0` and `1`.

🗣️

Speaker labels are not identities. Labels tell voices apart within one recording. They don't identify an employee, customer, extension, or named person. Utterances are returned as stored, so their fields may vary by transcription provider.

Example response

This example is illustrative. Status values and summary properties depend on your processing configuration and stored data.

JSON

{
  "xml_cdr_uuid": "c0ec8113-aa15-40ac-8437-47185dd9dcf4",
  "status": "completed",
  "summary_status": "completed",
  "created_at": "2026-10-19 20:15:30",
  "summary_payload": {
    "summary": "The caller asked about vehicle body repairs and was referred to a collision repair center.",
    "sentiment": "neutral",
    "action_items": [
      "Contact the recommended collision repair center."
    ]
  },
  "utterances": [
    {
      "end": 110830,
      "text": "Sorry, you said Sterling Collision Center?",
      "start": 107710,
      "speaker": "B",
      "confidence": 0.8814426
    },
    {
      "end": 112990,
      "text": "Correct.",
      "start": 112350,
      "speaker": "C",
      "confidence": 0.9996745
    }
  ]
}

Processing Behavior

Transcription and summarization run separately. A successful response means a transcription record was found, not that both stages have finished.

summary_payload may be null while the summary is unavailable.
utterances may be null while the conversation segments are unavailable.
An empty utterances array means the stored transcript has no segments.
Check both status and summary_status to tell whether processing has finished.

⏳

Polling. If you're waiting on processing, retry with a reasonable delay and backoff. Avoid rapid continuous polling, and stop retrying once processing reports a terminal failure.

Errors

HTTP status	Meaning
`400`	An identifier failed UUID validation (after authorization checks).
`401`	Authentication is missing or invalid.
`403`	The API user lacks access to the domain or the required permission.
`404`	The call wasn't found in the requested domain, or it has no transcription record.
`429`	The API rate limit was exceeded.
`500`	An unexpected server error occurred.

A call with no transcription record returns the error code transcription_not_found. That doesn't necessarily mean processing failed: the record may not have been created yet, or transcription may not have been requested for the call.

Authorization checks run before identifier validation, so a request for a domain you can't access may return 403 before its UUIDs are checked.

Integration notes

Use the xml_cdr_uuid from List Call History to match analysis to the right call in your application.
This endpoint only retrieves existing analysis. It doesn't start transcription, regenerate summaries, or return audio. For audio, use Recording URL.

🔒

Treat transcripts as private customer data. Responses include Cache-Control: private, no-store. Restrict access to transcripts and summaries within your application, and never expose your API token in public website code.