Hacker News Reader

Comment by torben-friis | Hacker News Reader

I recently used some subtitles that I later found out had been AI generated.

The experience wasn't really good to be honest - the text was technically correct but the way multiline phrases were split made them somehow extremely hard to parse. The only reason I learned AI was involved is that it was bad enough for me to stop viewing and check what was wrong.

Hopefully it's an implementation detail and we'll see better options in the future, as finding subtitles for foreign shows is always a pain.

View on HN

This reminds me of Prime Video subtitles. Anything not Hollywood blockbuster will only have one language (from what it looks like, randomly chosen) of garbage quality (not sure whether AI generated though). But there's worse anyway - some Asian titles ONLY available in badly dubbed versions - again in some random language (hello Casshern in... German???). So I see this VLC initiative as an improvement from this very very low bar.

Proper subtitles are obviously better, but it's impossible to do on everything. The tech is going to get better, and is already a game changer for hearing impaired people. Subtitles that are mostly correct are much better than none at all.

VLC has the option to find subtitles. If you use plex or jellyfin there are add-ons or bazarr, which does it automatically

On Youtube I really like the auto-generated captions and often prefer them to the creator's because:

- Sometimes the creator bases their captions on the script and misses changes in edit

- Sometimes the creator's captions are perfect transcriptions but broken up and timed awkwardly

Auto-generated captions aren't always perfect but unlike human captions provide word-by-word timing.

Yeah, I definitely wish other media would experiment with Youtube-style word-at-a-time subtitles. They often feel a lot more natural than full-sentence subtitles, the way they stream in is better at providing "connecting tissue", they never spoil upcoming reveals the way subtitles tend to, etc.

(By "connecting tissue", I mean they don't have the problem where sentence A is "I like chocolate", sentence B is "only on special occasions", and at the time B appears A is completely gone, but you really need A and B to be onscreen at the same time to parse the full meaning intuitively.)

As a Romanian, I'm so sick of AI translations on YouTube, especially since they use Google's translation (OpenAI's at least works quite well). Here's an example (translated back to English):

> Man Builds Background of Tan for Swedish House

It's completely puzzling. To understand it you have to know both English and Romanian. "Background of tan" is the term for "foundation" (makeup) in Romanian. That is, "foundation" has two meanings in English: for a house and for makeup, but Google uses the wrong one.

Automatic translation is full of these bad associations. I have no idea how people who don't speak English understand these translations.

I had exactly the same experience. Human expertise in subtitle makes the experience really better

If you have the luxury of requiring subtitles in English, sure. There's a huge scene of people making them and high quality subtitles available for pretty much everything. If you need subs in another language though your experience might change dramatically. Especially for any media that is old or less popular, in which case your options are probably either really bad subs, out of sync subs, or most likely, none whatsoever.

Reply on news.ycombinator.com