tech.lgbt is one of the many independent Mastodon servers you can use to participate in the fediverse.
We welcome all marginalized identities. This Mastodon instance is generally for folks who are LGBTQIA+ and Allies with an interest in tech work, academics, or technology in general.

Server stats:

2.9K
active users

Public

VLC jumps onto an AI bandwagon. Any recommendations for a cross platform player with a decent UI?

Source: techcrunch.com/2025/01/09/vlc-

Edit: okay, this ended up being inflammatory. This post was meant to only ask for player recommendations.

Yes, they're using STT and call it AI. They seem to be using open-source OpenAI Whisper trained on closed-sourced (potentially unethically sourced) data. Yes, it is an optional feature. Yes, STT is a useful application of machine learning. No, not all people who need subtitles are happy with auto-generated subtitles. Thank you, thank you, I love you all, too. 🥰

TechCrunch · VLC tops 6 billion downloads, previews AI-generated subtitles | TechCrunchVLC media player, the popular open source software developed by nonprofit VideoLAN, has topped 6 billion downloads worldwide and teased an AI-powered VLC media player, the open-source video software developed by nonprofit VideoLan, has topped 6 billion downloads.
Quiet public

@nina_kali_nina GodDAMMIT

He tells corpos to kick rocks for decades and then does THIS?

Fuuuuck, man...

Quiet public

@nina_kali_nina @TeflonTrout does it? All I can see is subtitles generated locally, can't find a mention of OpenAI from a quick search

Quiet public

@null @TeflonTrout there's a discussion on the orange site mentioning that it's OpenAI Whisper under the hood. Even if it was something different, I wouldn't be too happy, though

Quiet public

@nina_kali_nina the article says open source AI models. Are you sure?

Quiet public

@null@ice.0xportal.social @nina_kali_nina@tech.lgbt @TeflonTrout@beige.party When I use "permissively licensed" in quotes, I mean that the model itself is MIT licensed as declared by OpenAI, but the data used to train it (such as a collection of TED talks) is absolutely not, and thus the effective licensing depends on your jurisdiction and how laws adapt in the future.

Situations of legal uncertainty are not new for VLC, though, given they maintain libdvdcss and libaacs...

Quiet public

@nina_kali_nina @TeflonTrout

Nope

"VLC automatic subtitles generation and translation based on local and open source AI models running on your machine working offline, and supporting numerous languages"

Local, open source, running on the user machine, working offline. That probably excludes Open AI.

Quiet public

@ParadeGrotesque we have checked the source code. We really did.

Quiet public

@ParadeGrotesque ikr :) but then they need A model...

Quiet public

@ParadeGrotesque @nina_kali_nina @TeflonTrout nope, uses the "bad" OpenAI Whisper local model rather than the "really fucking bad" OpenAI GPT models on the cloud

Quiet public

@ParadeGrotesque @nina_kali_nina @TeflonTrout
For subtitles?! 😲
I hoped it would at least be something, you know… useful, like neural network-based upscaling. Goddammit, indeed — why?!!!