Manipulate transcript

What are you building here

In this example, we are going to manipulate a bit the transcript before saving it to a file.

Please refer to previous example regarding authentication, pagination, and how to run such scripts.

Let's say that for some business purpose, we want to create a CSV file, one call per row, and display the unique identifier of the Call, a link to Scribr, our speech analytics tool, and the first 50 words a customer said during the Call.

Code

In this example, we're going to stop consuming the API after 3 pages. For more information about the resources of the API, please check the reference.

#!/bin/env python3
import csv
import os
import requests

activate_token = os.getenv('ACTIVATE_API_TOKEN')
activate_url = os.getenv('ACTIVATE_API_URL')
headers = {"Authorization": f"Token {activate_token}"}
LIMIT = 20
offset = 0


# First get the 3 pages we want and store them in a variable
call_list = []
while offset < 3 * LIMIT:
    r = requests.get(f"{activate_url}/calls?limit={LIMIT}&offset={offset}", headers=headers)
    r.raise_for_status()
    data = r.json()
    call_list += data["data"]
    offset += LIMIT

output = []
for call in call_list:
    # for each call, transcript is within key `transcript_json`. Aside from metadata, `callData` is the key
    # you're looking for for the transcript itself.
    transcript = call["transcript_json"].get("callData", [])
    words = []
    for segment in transcript:
        # A transcript is made of segments: utterance of a given speaker.
        # We're just going to keep only the customer's, named `in`.
        if segment["from"] == "in":
            words += [word["value"] for word in segment["words"]]

    # And now we can prepare the data we'll save in CSV:
    # - call unique identifier
    # - a link to our website
    # - 50 first words of the customer
    output.append({
        'unique_id': call['unique_id'],
        'url': f"https://app.uh.live/scribr/{call['unique_id']}",
        '50words': " ".join(words[:50])
    })

with open("client_first_words.csv", "w") as csvfile:
    writer = csv.DictWriter(csvfile, fieldnames=['unique_id', 'url', '50words'])
    writer.writeheader()
    writer.writerows(output)