Very slow extraction of hardcoded subtitles with VideOCR

[frenksisco]

Hi, I'm trying to extract hardcoded subtitles with VideOCR but the operation is extremely slow even though I have GPU usage enabled. The GPU is working at almost 0%... Can anyone tell me why?
I installed the “VideOCR-GPU-v1.3.2-CUDA-12.9-setup-x64.exe” version since it is recommended for Nvidia 16 - 50 Series cards and I have an Nvidia RTX 2070 Super GPU.

I also get the error: Error: PaddleOCR failed. See the log file for technical details
The log says:

[2025-10-07 11:43:30] PaddleOCR process failed with exit code 3221225794.
Command: C:\PROGRA~1\VideOCR\videocr-cli-GPU-v1.3.2-CUDA-12.9\PaddleOCR-GPU-v1.3.2-CUDA-12.9\paddleocr.exe ocr --input C:\Users\frenk\AppData\Local\Temp\videocr_temp_afh o7ewe --device gpu --use_textline_orientation false --use_doc_orientation_classify False --use_doc_unwarping False --lang it --text_detection_model_dir C:\PROGRA~1\VideOCR\videocr-cli-GPU-v1.3.2-CUDA-12.9\PaddleOCR.PP-OCRv5.support.files\det\PP-OCRv5_mobile_det --text_detection_model_name PP-OCRv5_mobile_det --text_recognition_model_dir C:\PROGRA~1\VideOCR\videocr-cli-GPU-v1.3.2-CUDA-12.9\PaddleOCR.PP-OCRv5.support.files\rec\latin_PP-OCRv5_mobile_rec --text_recognition_model_name latin_PP-OCRv5_mobile_rec

--- STDOUT ---


--- STDERR ---

[Subtitles]

From experience hard coded subtitles are very hard to extract
Sorry I am not familiar with VideOCR nor would I try it.
You can save time if you try to transcribe the audio from the video (assuming it has an audio stream) using Whisper AI
Or you can try to follow this YouTube video.
https://www.youtube.com/watch?v=bgg3otPw810