The first concept to understand is that of a job.

A job represents a specific task initiated by the user, just as summarizing a call or extracting questions and answers from a transcript. A job is characterized by parameters and pipelines passed to the API.

A job can be created by two tasks: Summarization or Extraction. The first is used to summarize a given transcript, while the second is used to extract useful information from a given transcript.

A job is created when the user initiates either a Summarization task or an Extraction task.

Summarization

Summarization is the task of extracting the most important information from an audio file or transcript and rewording it to provide a user with a much shorter, more readable version of the original source material.

Note that the Wordcab API only generates abstractive summaries. To start a summarization job, you need to send a POST request to the /summarize endpoint (see Start Summary).

Extraction

Extraction is the process of extracting useful information from a given transcript. You can extract the following information:

  • Emotions: The emotions detected in the transcript for each utterance.
  • Topic changes: The transcript points where a topic change occurs.
  • Questions/Answers: The questions and answers detected in the transcript.
  • Talk ratios: The talk ratios detected in the transcript for each speaker.

To start an extraction job, you need to send a POST request to the /extract endpoint (see Start Extract).


What’s Next

We will talk about Transcripts.