Use whisper to detect shows previews

[Hello-World-Traveler]

I am not sure if this is possible for this plugin.

The user could host whisper on docker; https://hub.docker.com/r/linuxserver/faster-whisper. The down side is that whisper is not 100% and uses more system resources.

This script may not work.

import requests
import moviepy.editor as mp
from pydub import AudioSegment
import io
import json

# Function to extract audio from video and get first 10 minutes
def extract_audio(video_file, duration=600):
    video = mp.VideoFileClip(video_file)
    audio = video.audio
    audio = audio.subclip(0, duration)
    audio.write_audiofile("temp_audio.wav")
    return "temp_audio.wav"

# Function to transcribe audio using remote Whisper API
def transcribe_audio(audio_file, api_url):
    with open(audio_file, "rb") as f:
        files = {"file": f}
        response = requests.post(api_url, files=files)
    return response.json()

# Function to process transcription and detect "previously"
def process_transcription(transcription):
    for segment in transcription["segments"]:
        if "previously" in segment["text"].lower():
            return segment["start"]
    return None

# Main script
def main():
    video_file = "path_to_your_45min_video.mp4"
    whisper_api_url = "https://api.whisper.example.com/transcribe"  # Replace with actual API endpoint

    # Extract first 10 minutes of audio
    audio_file = extract_audio(video_file)

    # Transcribe audio using remote Whisper API
    transcription = transcribe_audio(audio_file, whisper_api_url)

    # Process transcription to find "previously"
    timestamp = process_transcription(transcription)

    if timestamp is not None:
        print(f"The word 'previously' was found at {timestamp} seconds")
    else:
        print("The word 'previously' was not found in the first 10 minutes")

if __name__ == "__main__":
    main()

[Hello-World-Traveler]

Or the plugin could scan subtitles

1
00:00:01,919 --> 00:00:03,889
Previously on.....

103
00:06:52,398 --> 00:06:55,408
Start of intro

111
00:07:13,143 --> 00:07:16,143
End of intro

[AbandonedCart]

Whisper is a bit much. The list of reasons why that wouldn't be ideal is also pretty long, so I'll leave it at just the main one ->
The plugin is meant to be self-contained inside Jellyfin and whisper is more for making subtitles than detecting intros.

[Hello-World-Traveler]

I do agree with you on that. It is a bit much.