Gather stats about oral argument audio [Deadline: EOD Monday]

[mlissner]

Describe the issue.

We need the following stats about our oral argument audio:

• How many hours: 55k
• How far back it goes: 1969
• How many words: ????

Related Issue

I got the hour-count and oldest item from the front end. Just need the word count.

Screenshots

No response

What type of issue is this?

Research

[ERosendo]

I ran the following Python script to calculate the total word count from our audio transcripts

from cl.audio.models import Audio
from django.contrib.humanize.templatetags.humanize import intword

total_words = 0
for audio in Audio.objects.all():
    if audio.stt_transcript:
        word_count = len(audio.stt_transcript.split())
        total_words += word_count

print(f"Total word count from stt_transcript field: {intword(total_words)}")

The script iterates through all Audio objects, sums the word count from the stt_transcript field of each, and formats the final count for readability. The script was executed in a maintenance pod, and the result was:

Total word count from stt_transcript field: 492.6 million

[ERosendo]

Here’s a refined version of the script. I realized some of our transcriptions are marked as hallucinations, so I excluded those from the word count:

from cl.audio.models import Audio
from django.contrib.humanize.templatetags.humanize import intword

total_words = 0
for audio in Audio.objects.filter(stt_status=Audio.STT_COMPLETE):
    word_count = len(audio.stt_transcript.split())
    total_words += word_count

print(f"Total word count from completed stt_transcript field: {intword(total_words)}")

Output:

Total word count from completed stt_transcript field: 491.6 million

[mlissner]

Thanks. Forwarded to client and posted on Bluesky: https://bsky.app/profile/free.law/post/3lydfbpa2uk2t