Interaction Analytics Export transcript formats

Each exported transcription file contains a number of JSON files, one for each transcribed interaction that is included in the export. The name of each JSON file corresponds to the call’s transaction ID. Each JSON file has three versions of the same transcription, each in a different format.

Open the JSON file in a JSON formatting application to display the content on the screen.

plainText

In this format, the whole transcription appears as a single sequence, without additional details. This format uses \n as a separator when the speaker changes.

For example:

example of plaintext - whole transcription as single sentence

plainTextTime

In this format, the transcription includes the details of the speaker and start and end time, and the sections are separated for each speaker.

For example:

Example of plain text transcript

transcript_detailed

In this format, each word is presented separately, with details such as the speaker, the word’s position, and start and end time.

For example:

Example of detailled transcription

Key name	Definition	Example	Meaning of the example
"p"	The index in which the word appears in the transcript. 1 is the first word, 2 is the second, and so on.	"p":14	This is the 14th word of the call
"c"	A value between 0 and 1 that indicates the confidence that this is the correct word	"c":0.88	There is 88% probability that this word is correct
"s"	Start time (in milliseconds)	"s":2200	The start time of this word was 2200 milliseconds from the start of the recording
"e"	End time (in milliseconds)	"e":2300	The end time of this word was 2300 milliseconds from the start of the recording
"w"	The transcribed word\words	"w":"Hello"	The transcribed word is "Hello"
"sp"	The speaker—for example, agent, customer, simultaneous	"sp":"Agent"	The speaker is the agent