- cross-posted to:
- [email protected]
- [email protected]
- a11y
- [email protected]
- cross-posted to:
- [email protected]
- [email protected]
- a11y
- [email protected]
cross-posted from: https://lemmy.ca/post/37011397
The popular open-source VLC video player was demonstrated on the floor of CES 2025 with automatic AI subtitling and translation, generated locally and offline in real time. Parent organization VideoLAN shared a video on Tuesday in which president Jean-Baptiste Kempf shows off the new feature, which uses open-source AI models to generate subtitles for videos in several languages.
You must log in or register to comment.
Finally, some good fucking AI
I was just thinking, this is exactly what AI should be used for. Pattern recognition, full stop.
Yup, and if it isn’t perfect that is ok as long as it is close enough.
Like getting name spellings wrong or mixing homophones is fine because it isn’t trying to be factually accurate.
Problem ist that now people will say that they don’t get to create accurate subtitles because VLC is doing the job for them.
Accessibility might suffer from that, because all subtitles are now just “good enough”
Or they can get OK ones with this tool, and fix the errors. Might save a lot of time
Regular old live broadcast closed captioning is pretty much ‘good enough’ and that is the standard I’m comparing to.
Actual subtitles created ahead of time should be perfect because they have the time to double check.
deleted by creator
From experience AI translation is still garbage, specially for languages like Chinese, Japanese, and Korean , but if it only subtitles in the actual language such creating English subtitles for English then it is probably fine.
That’s probably more due to lack of training than anything else. Existing models are mostly made by American companies and trained on English-language material. Naturally, the further you get from the model, the worse the result.
It is not the lack of training material that is the issue, it doesn’t understand context and cultural references. Someone commented here that crunchyroll AI subtitles translated Asura Hall a name to asshole.
deleted by creator
I have a feeling that if you care enough about subtitles you’re going to look for good ones, instead of using “ok” ai subs.
I imagine it would be not-exactly-simple-but-not- complicated to add a “threshold” feature. If Ai is less than X% certain, it can request human clarification.
Edit: Derp. I forgot about the “real time” part. Still, as others have said, even a single botched word would still work well enough with context.
That defeats the purpose of doing it in real time as it would introduce a delay.
Derp. You’re right, I’ve added an edit to my comment.
I’d like to see this fix the most annoying part about subtitles, timing. find transcript/any subs on the Internet and have the AI align it with the audio properly.
YES! I can’t stand when subtitles are misaligned to the video. If this AI tool could help with that, it would be super useful.
deleted by creator
Finally some good AI fucking 🤭
What’s important is that this is running on your machine locally, offline, without any cloud services. It runs directly inside the executable
YES, thank you JB
Justin Bieber?
JB stands for Jean-Baptiste, who is the main maintainer of vlc
Good ol’ Zorg
Jack Black?
James Brown?
❤️
This sounds like a great thing for deaf people and just in general, but I don’t think AI will ever replace anime fansub makers who have no problem throwing a wall of text on screen for a split second just to explain an obscure untranslatable pun.
Bless those subbers. I love those walls of text.
Translator’s note: keikaku means plan
They are like the * in any Terry Pratchett (GNU) novel, sometimes a funny joke can have a little more spice added to make it even funnier
It’s unlikely to even replace good subtitles, fan or not. It’s just a nice thing to have for a lot of content though.
I have family members who can’t really understand spoken English because it’s a bit fast, and can’t read English subtitles again, because again, too fast for them.
Sometimes you download a movie and all the Estonian subtitles are for an older release and they desynchronize. Sometimes you can barely even find synchronized English subtitles, so even that doesn’t work.
This seems like a godsend, honestly.
Funnily enough, of all the streaming services, I’m again going to have to commend Apple TV+ here. Their shit has Estonian subtitles. Netflix, Prime, etc, do not. Meaning if I’m watching with a family member who doesn’t understand English well, I’ll watch Apple TV+ with a subscription, and everything else is going to be pirated for subtitles. So I don’t bother subscribing anymore. We’re a tiny country, but for some reason Apple of all companies has chosen to acknowledge us. Meanwhile, I was setting up an Xbox for someone a few years ago, and Estonia just… straight up doesn’t exist. I’m not talking about language support - you literally couldn’t pick it as your LOCATION.
For all their faults, Apple knows accessibility. Good job Timmy.
That still happens? Maybe wanna share your groups? ;)
Now I want some AR glasses that display subtitles above someone’s head when they talk à la Cyberpunk that also auto-translates. Of course, it has to be done entirely locally.
I guess we have most of the ingredients to make this happen. Software-wise we’re there, hardware wise I’m still waiting for AR glasses I can replace my normal glasses with (that I wear 24/7 except for sleep). I’d accept having to carry a spare in a charging case so I swap them out once a day or something but other than that I want them to be close enough in terms of weight and comfort to my regular glasses and just give me AR like overlaid GPS, notifications, etc, and indeed instant translation with subtitles would be a function that I could see having a massive impact on civilization tbh.
I believe you can put prescription lenses in most AR glasses out there, but I suppose the battery is a concern…
I’m in the same boat, I gotta wear my glasses 24/7.
I think we’re closer with hardware than software. the xreal/rokid category of hmds are comfortable enough to wear all day, and I don’t mind a cable running from behind my ear under a clothes layer to a phone or mini PC in my pocket. Unfortunately you still need to byo cameras to get the overlays appearing in the correct points in space, but cameras are cheap, I suspect these glasses will grow some cameras in the next couple of iterations.
It’d be incredible for deaf people being able to read captions for spoken conversations and to have the other person’s glasses translate from ASL to English.
Honestly I’d be a bit shocked if the AI ASL -> English doesn’t exist already, there’s so much training data available, the Deaf community loves video for obvious reasons.
soon
Breaking news: “WW3 starts over an insult due to a mistranslated phrase at the G7 summit. We will be nuked in 37 seconds. Fuck like rabbits, it’s all we can do. Now over to Robert with traffic.”
There are plenty mistranslations in politics done by humans already.
As vlc is open source, can we expect this technology to also be available for, say, jellyfin, so that I can for once and for all have subtitles.done right?
Edit: I think it’s great that vlc has this, but this sounds like something many other apps could benefit from
It’s already available for anyone to use. https://github.com/openai/whisper
They’re using OpenAI’s Whisper model for this: https://code.videolan.org/videolan/vlc/-/merge_requests/5155
Has there been any estimated minimal system requirements for this yet, since it runs locally?
It’s actually using whisper.cpp
From the README:
Memory usage Model Disk Mem tiny 75 MiB ~273 MB base 142 MiB ~388 MB small 466 MiB ~852 MB medium 1.5 GiB ~2.1 GB large 2.9 GiB ~3.9 GiB
Those are the model sizes
Oh wow those pretty tiny memory requirements for a decent modern system! That’s actually very impressive! :D
Many people can probably even run this on older media servers or even just a plain NAS! That’s awesome! :D
Note that openai’s original whisper models are pretty slow; in my experience the distil-whisper project (via a tool like whisperx) is more than 10x faster.
deleted by creator
That explains why their subtitles have seemed worse to me lately. Every now and then I see something obviously wrong and wonder how it got by anyone who looked at it. Now I know why. No one looked at it.
deleted by creator
Malevolent Kitchen Intensifies
I hope it’s available for Stash App. I wanna know what this JAV girls are saying.
( ͡° ͜ʖ ͡°)
In the *arr suite, bazarr has a plugin called Subgen which you can add and you can set it to generate subtitles on your entire library if you want, or only missing subtitles. The sync is spot on compared to 90% of what Opensubtitles delivers. I sometimes re-gen them with this plugin just because opensubtitles is so constantly out of sync (e.g. highly rated subtitles 4 lines will be at breakneck pace and the next 10 will be super slow and then everything is 3 seconds off)
It isn’t in-player but it works. The downside is it is a larger model and takes ~20 minutes to generate a movie length of subtitles.
Ooooh I like this
This might be one of the few times I’ve seen AI being useful and not just slapped on something for marketing purposes.
And not to do evil shit
But the toppings contains potassium benzoate.
As long as the models are OpenSource I have no complains
And the data stays local.
And yet they turned down having thumbnails for seeking because it would be too resource intensive. 😐
I mean, it would. For example Jellyfin implements it, but it does so by extracting the pictures ahead of time and saving them. It takes days to do this for my library.
Yeah, I do this for plex as well, and stash. I think if the file already exists in the directory vlc should use it. It’s up to you to generate them. That is exactly how cover art for albums on songs worked in VLC for a decade before they added the feature to pull cover art on the fly.
I get what you are saying, but I don’t think there is any standardized format for these trickplay images. The same images from Plex would likely not be usable in Jellyfin without converting the metadata (e.g. to which time in the video an image belongs to). So VLC probably does not have a good way to understand trickplay images not made by VLC.
Video decoding is resource intensive. We’re used to it, we have hardware acceleration for some of it, but spewing something around 52 million pixels every second from a highly compressed data source is not cheap. I’m not sure how both compare, but small LLM models are not that costly to run if you don’t factor their creation in.
All they’d need to do is generate thumbnails for every period on video load. Make that period adjustable. Might take a few extra seconds to load a video. Make it off by default if they’re worried about the performance hit.
There are other desktop video players that make this work.
It is useful for internet streams though, not really for local or lan video.
I hope Mozilla can benefit of a good local translation engine that could come out of it as well.
They technically already do with Project Bergamot.
I know they do, but it’s lacking so many languages.
And it takes forever. I’m using the TWP plugin for Firefox (which uses external resources, configurable to google, bing and yandex translate respectively), and it’s near instantaneous. The local one from Mozilla often takes 30 seconds, and sometimes hangs until I refresh the page.
Weird, I can translate with FF even on my phone and it’s not a strong one.
The nice thing is, now at least this can be used with live tv from other countries and languages.
Think you want to watch Japanese tv or Korean channels with out bothering about downloading, searching and syncing subtitles
I prefer watching Mexican football announcers, and it would be nice to know what they’re saying. Though that might actually detract from the experience.
GOOOOOOAAAAAAAAALLLLLLLLLL
Just fill up the whole screen with this.
The opposing team has scored.
Amazing. I can finally find out exactly what that nurse is yelling about while she gets railed by the local basketball team.
Something about a full-court press?
Will it be possible to export these AI subs?
Imagine the possibilities!
The technology is nowhere near being good though. On synthetic tests, on the data it was trained and tweeked on, maybe, I don’t know.
I corun an event when we invite speakers from all over the world, and we tried every way to generate subtitles, all of them run on the level of YouTube autogenerated ones. It’s better than nothing, but you can’t rely on it really.No, but I think it would be super helpful to synchronize subtitles that are not aligned to the video.
This is already trivial. Bazarr has been doing it for all my subtitles for almost a decade.
is your goal to rely on it, or to have it as a backup?
For my purpose of having backup nearly anything will be better than nothing.When you do live streaming there is no time for backup, it either works or not. Better than nothing, that’s for sure, but also maybe marginally better than whatever we had 10 years ago
Really? This is the opposite of my experience with (distil-)whisper - I use it to generate subtitles for stuff like podcasts and was stunned at first by how high-quality the results are. I typically use distil-whisper/distil-large-v3, locally. Was it among the models you tried?
I unfortunately don’t know the specific names of the models, I will comment additionally if I will not forget to ask people who spun up the models themselves.
The difference might be that live vs recorded stuff, I don’t know.
You were not able to test it yet calling it nowhere near good 🤦🏻
Like how should you know?!
Relax, they didn’t write a new way of doing magic, they integrated a solution from the market.
I don’t know what the new BMW car they introduce this year is capable of, but I know for a fact it can’t fly.
Haven’t watched the video yet, but it makes a lot of sense that you could train an AI using already subtitled movies and their audio. There are times when official subtitles paraphrase the speech to make it easier to read quickly, so I wonder how that would work. There’s also just a lot of voice recognition everywhere nowadays, so maybe that’s all they need?
When does this get released? I really want to try it
