Are you interested in a tool that extracts burnt-in subtitles from videos?

Hi LingQers,

I am a software developer who also loves learning Chinese. One of the biggest challenges in learning Chinese is the fact that it is difficult to find content with both audio and subtitles.

Oftentimes videos in Chinese will have subtitles, but they are usually burnt in to the video itself so there is no way to easily extract them.

However, I recently hacked together a tool that allows me to extract burnt-in subtitles from any video (in pretty much any language).

I want to see how much interest there is in the market for a tool like this. If the interest is enough, I might build it out into a fully-fledged product where anyone can upload a video and in a few minutes have its subtitles.

Could you please answer the following questions for me?

  1. How interested are you in a product like this?
  2. If it existed, what features would you want it to have?
  3. What price range seems fair to you?

Proof of concept demo (yes, it actually works):

The actual product would not be code-based, it would be a website like LingQ that is easy to use. This is just a demo of my hack.

I pulled from 20s to 30s in this video:

Result:

9 Likes

I myself don’t have a use for it, but like that someone like you is out there floating this idea.

Problem–>Market–>Solution–>Progress of Mankind.

I love capitalism.

2 Likes

The feature sounds great, but how does it work? Because if you’re using a programme for it, why would paying per hour make sense for users instead of paying for the product once or having monthly/ annual fees like other products do?

1 Like

The you see above is using Google’s Vision text detection API, so it actually does have a cost associated with running it.

Yes, that cost is small, but just selling the software by itself without a cost for each run-through would not be a viable business model.

However, if there is enough interest in something like this and people really want it to be a one-time cost instead, I’d consider building my own OCR algorithm. However, I’d have a lot of learning to do before I could build something like this lol.

2 Likes

This would be amazing - I think it’s a feature people would pay for. The price you mentioned is a bit steep though.

  1. Very interested
  2. If it can rip both hardcoded subtitles and audio - that would be amazing.
  3. Bit pricey.
1 Like

What do you think would be a fair price in accordance with my payment model (per hour)?

Also happy to hear other payment ideas, but that’s my main goto right now.

That’s a great idea, but there are other free products on the market that offer similar functionality. There is open source OCR available. I hate to pay for the hour, even if yours is really easy to use.

1 Like

Can you point me to where those free products are?

I’ve seen AVISubDetector, VideoSubFinder, and a few others but all of them are:

  1. Difficult to use
  2. Time consuming
  3. Not adaptable to many languages
    If you’d hate to pay for the hour, what would be a fair price or payment model for you?
1 Like

This would be very very useful for Chinese learners. It’s odd that nearly all Chinese programs have subtitles but none of them seem to use close captioning or SRT formats. Very frustrating.

If your program could also strip the audio and save that, you’d save everyone a second step and have the perfect tool for Chinese LingQ! One other suggestion is to have the option to generate time stamps for each line.

I would pay for this on a subscription basis (maybe even $10 per month) but $7 per hour is way too steep. For that price you could hire someone in China to sit and transcribe the characters off the video.

If you do implement an open source OCR based version (and therefore bring the cost down), please let me know. I’ll be your first customer.

3 Likes

Edwin, if you haven’t already done so, it may be worth floating the idea over at https://www.chinese-forums.com/

You’ll get the most sophisticated feedback over there. And it is how a number of interesting products get off the ground in the “foreigners learning Chinese” market.

fwiw, a lot of us old timers have spent too many years pausing programs and ocr’ing obscure characters, to not see the instant value in this.

2 Likes
  1. How interested are you in a product like this? I would like to use this for Indonesian. it is hard to find CC for indonesian.
  2. If it existed, what features would you want it to have? same as others have mentioned.
  3. What price range seems fair to you? monthly or annual subscription but not per use…
1 Like

I have been wanting to improve my Hebrew for the past 6 years, but literally ever tv show and channel has the subtitles embedded into the video, so there’s no way for me to get it in to LingQ. If I could extract the text it would be a game changer completely.

Yep, all those programs suck, but because they exist your product wouldn’t be that unique to justify the price. I’ve dabbled with the copyfish addon for Firefox (it has its issues as well). I should clarify that the reasons I don’t like the hourly model are that A)paying=pain B)you encourage your users to use your program as little as possible. Freemium (like Pleco) is best, second best is a one time payment, third a subscription. Isn’t there open source OCR? https://www.quora.com/What-are-the-best-open-source-OCR-libraries
videocr · PyPI