using this new transcription method and it seems to catch most of the conversation properly ?
I don’t have any of the films you’ve poster do I haven’t look at subtitles, I sang curious about you process, Care to share your technique?Here is my another test using this new transcription method and it seems to catch most of the conversation properly and the plot is 90% understandable after translating though DeepL. I will share the English sub here and would be glad to know peoples feedback if the sub are understandable for them or not.
HUNTA-371
I am still testing at the moment and will provide teh method after simplifying the steps to make it work in future. So everyone can use it easily.
Feedback:Here is my another test using this new transcription method and it seems to catch most of the conversation properly and the plot is 90% understandable after translating though DeepL. I will share the English sub here and would be glad to know peoples feedback if the sub are understandable for them or not.
HUNTA-371
It takes around 1 hour for a 3hr long movie and these subtitle are autogenerated by audio transcription in Japanese. This method is different from autosub and pytranscribe as it doesnt use Google deep speech for generating sub.
Ok so I'm checking out hunta 666A.
Thanks for the feedback and also I am making these subs for myself as I was not able to find subs of most hunta series AV I want to jerk off too. So I decided to make the subs myself and share it with you guys.Ok so I'm checking out hunta 666A.
I added images of what I seen.
Assessment:
1.There are mistakes.
2. It lacks basic sex words used a lot in JAV, for examples, dick(cock), pussy(cunt), sugoi, kimochii, etc. I find their absences to be "weird". But I don't know
Feedback:
1. Slow down and perfect your method. Don't try to do so many at once.
2. I think it's capturing some good dialog.
3. Talking about #2 in the Assessment, it's possible I seen some JAV not used cock or pussy.... but "sugoi or kimochii" are usually there doing a sex scene?
Admittedly, I could be super wrong. Not sure if r*** jav have those words?
My final take:
Your method is still good. Everyone should always double check the subs anyways
Yes, I understand. That's why I started translating as well lol
The most useful thing about video sub Finder is that it finds the subtitles, cuts them out and creates a folder called RGB-Images that contains Jpeg image of each subtitle named by the time it appears, don’t use the ocr Billy into vsf it’s not good, take that RGB-image folder and proceeds it with another OCR program to populate the Text/result folder with OCRed text, then use VSF again to turn the OCRed text into subtitles.
thank you for guiding .
in my opinion and because of my laptop spec on windows 8, the https://github.com/Abu3safeer/image-ocr-google-docs-srt is more reliable than easyocr, using the same RGB-Images from videosubfinder, but the best one is "Google Lens apk" the cons only available using smartphone, if only someone can make exe or py version it would be great.
