rqst: OCR for subtitles

[iuvi7]

Hello!

Is it possible to AI recognize subtitles that "sewed" into video? To extract text and create separate file? with timelines and text of course

If not - maybe in plans? Never saw any kind of AI that could do such cherry thing

Thanks

[felipe-lima-coelho]

Wouldn't it be easier to use speech-to-text? For example, whisper or something similar? I say this because OCR from a video tends to be very expensive.

[iuvi7]

You don't get me right. There no sound or speech at all in video. Just subtitles

[felipe-lima-coelho]

You don't get me right. There no sound or speech at all in video. Just subtitles

With this new context, everything changes.