This API submits a call recording for transcription. The transcript is written back into the conversation as speaker-attributed messages.
The request is accepted immediately and processed asynchronously – use the returned request_id with the GET status API to follow its progress.
The recording can be supplied in one of three ways:
- activity_id – reference an existing call-recording activity already on the conversation.
- recording_path – a path on the channel's configured recording domain, which CommBox downloads.
- file upload – send the audio file directly as multipart/form-data.
Required path param is the Object (conversation) Id.
(POST) /objects/OBJECT_ID/transcribe
Example:
https://api.commbox.io/objects/293854/transcribe
API key is needed to authorize requests. You can get your API key via the management console.
A Bearer Token is needed to be set in the Authorization header of every API call.
For additional support you can contact us.
Object Id is a CommBox generated number for the conversation.
Use application/json to reference an existing activity (activity_id) or a recording path (recording_path). Use multipart/form-data to upload the audio file itself.
A CommBox generated number for an existing call-recording activity on the conversation. Mutually exclusive with recording_path / file upload.
Path of the recording on the channel's configured recording domain (relative, must not be an absolute URL). Mutually exclusive with activity_id.
Your own identifier for this request. When supplied, the request becomes idempotent – repeating it with identical content returns the original request_id.
Ordered list of the speakers expected in the recording. An entry with managerId is attributed to that agent; an entry without one is attributed to the customer of the conversation.
Display name of the speaker.
A CommBox generated number for the agent (manager). Omit for the customer.
The audio file to transcribe.
Your own identifier for this request (idempotency key).
Alternative to uploading a file – a path on the channel's configured recording domain.
The speakers array, JSON-encoded as a string.
Accepted - The transcription request was accepted and is being processed
A CommBox generated identifier for the transcription request. Use it with the GET status API.
Status of the request at submission time.
A CommBox generated number for the conversation.
The referenced call-recording activity. Returned only when activity_id was supplied – otherwise the activity is created during processing and appears in the GET status response.
Echo of the supplied client_reference_id. Returned only when one was supplied.
Bad Request - The request could not be understood by the server. Incoming parameters might not be valid (missing_input, ambiguous_input, invalid_speakers, activity_id_required, activity_not_found, activity_wrong_type, file_too_large, blocked_url)
Machine-readable error code.
Unauthorized - The supplied credentials, if any, are not sufficient to access the resource
Not found - The requested resource is not found
Conflict – The provided client_reference_id has already been used for a different transcription request. Use a new client_reference_id or resend the original request content
Unprocessable Entity - Transcription is not installed or not enabled for this channel (transcription_not_configured)
Too Many Requests - Too many concurrent transcription requests for this brand (Throttling). A Retry-After header is returned
Server Error - The server could not return the representation due to an internal server error
Not Implemented - The requested operation is not supported