Object Structures

A typical response of a specific call from our Speech Analytics API contains the following object properties.

CDRs

CDR object contains basic details about the call.

Property Type Description Example
requestUri string Connected voice URI, destination sip:fikrippi@ippi.fr
relatedVoxboneCallId string Unique identifier for the call NFHP7QXJ5VBNRGNG5TEDRK4Y6Y
callingNumber string The calling number, “from:” address 32493400606
calledNumber string The called number, “To:” address  883510080983
startTime string  The date-time timestamp in UTC for when the call has started 2017-11-17T13:38:00.000Z
connectTime string  The date-time timestamp in UTC for when the call got connected 2017-11-17T13:38:01.000Z
endTime string  The date-time timestamp in UTC for when the call was ended 2017-11-17T13:40:56.000Z
duration integer  The duration of the call in seconds, at int32 format 55

 

Recordings

Contains recorded file information related to a specific call along with links to the file you can use to download or play the recording.

Property Type Description Example
id integer Unique identifier of the recording, in int64 format 1
callId string Internal call ID generated by the recording system 86456775-5b7c-47c0-8e7c-3089f1186c2f
downloadLink string Download link for the recorded audio file https://../callee-NFHP7QXJ5VBNRGNG5TED4Y6Y@81.201.82.106.flac?generation=1510925968503714&alt=media
format string Recording file format flac
recorder string Recording system used custom-recording-service
codec string The audio codec that was used in media transmission PCMA
duration integer Duration of the recorded file in seconds, in Int32 format 50
channels integer Number of channels in the recording in Int32 format 1
createdAt string The date-time timestamp of the recording session creation  2017-11-17T13:38:00.000Z
endedAt  string  The date-time timestamp of the recording session completion  2017-11-17T13:40:56.000Z
participantDescription  string enum string description of the participant related to the recorded file  – caller
– callee
completion  string enum  completion status of the recording session  – normal
– failed
recordingProfileId  string  Voxbone recording/speech analytics profile Id as unique identifier  741a3e3e-0677-4655-a891-02457800973d
updatedAt  string Date-time timestamp of the last update performed on the recordings object  2017-11-17T13:40:56.000Z

 

Analytics

Analytics object contains all results from speech analytics performed on the call leg such as transcription, talk time analysis and sentiment scores.

Talk Time Analysis

Shows the ratio of talk time of participant vs. the full duration of the call. Also shows how much the participant talked over the other participant (overtalk). Has value between 0 and 1.

 

Property Type Description Example
overTalk number (double) Overtalk amount for this participant over the other. Ranges from 0 to 1 0.05
talkTime number (double) Portion of the call where this participant spoke 0.64
failureReason string If the computation failed, reason for the failure Second call leg missing

 

Transcription

Contains speech to text transcription of the recorded participant in words with timestamps in miliseconds.

 

Property Type Description Example
speechToText Array (Object)  Array containing speech to text data
transcript Array(Object) Array containing word and timestamp collection
speechToText.transcript[]  startTimestamp integer (INT32) Start timestamp for this word, in miliseconds 85530
speechToText.transcript[] endTimestamp integer (INT32) End timestamp for this word, in miliseconds 86650
speechToText.transcript[] word string Transcribed word Hello
plainTranscription string Transcription in plain text for this call leg. Searchable field Hello and welcome to Voxbone
lang string Language selected for this transcription. If present, overwrites the
global language for this service, selected in Analytics settings. This
is the language selected per DID basis
en-US
failureReason string Failure reason for the speech to text api call Unsupported language code

 

Sentiment Analysis

Shows sentiment score per sentence for the recorded participants. Values range between -1 and 1, showing negative and positive sentiments.

Property Type Description Example
score number (double) Sentiment score between -1.0 (negative sentiment) and 1.0 (positive sentiment) for the whole document 0.4
magnitude number (double) A non-negative number in the [0, +inf) range, which represents the absolute magnitude of sentiment regardless of score (positive or negative) for the whole document 0.5
language string Autodetected language by the sentiment analysis engine en
failureReason string Description of failure Unsupported language code
sentences Array(object) Array of all sentences present in the call that sentiment analysis has been performed  
sentences[] sentiment Object Magnitude and score collection of each sentence’s sentiment score as described above for overall values – magnitude : 0.5
– score : 0.2
sentences[] text Object Sentences that sentiment analysis has been performed on  
sentences[] text.content String A single sentence from the whole transcript that’s the reference for the magnitude and scores of the sentences object item Hello and welcome to Voxbone.
beginOffset integer(INT32) The API calculates the beginning offset of the content in
the original document according to the EncodingType
specified in the API request
85530