Futility Closet: "Alienated Majesty"

https://www.futilitycloset.com/2025/03/22/alienated-majesty/

Preventing AI Plagiarism With .ASS Subtitling - Around two years ago, the world was inundated with news about how generative AI or... - https://hackaday.com/2025/01/25/preventing-ai-plagiarism-with-ass/ #artificialintelligence #generativeai #plagiarism #subtitles #youtube #poison #ass #srt #ai

Here @razzlom share his wisdom! ​​

If you have video file like this (pic 1) and you want video file like that (pic 2) just run this command:

ffmpeg -i INPUT.mkv -map 0:v -c:v copy -disposition:v default -map 0:a:m:language:jpn -c:a libopus -b:a 128k -q:a 0 -disposition:a default -map 0:s:m:language:rus -c:s copy -disposition:s default -map 0:t -c:t copy OUTPUT.mkv

It's really great that some YouTubers go to the effort of transcribing the video and set the timing for the subtitles.

But then they EMBED THEM IN THE VIDEO instead of uploading them directly to YouTube, which means they aren't very accessible and you're unable to translate them.

Urgh, it pains me that so many #YouTube creators and even bigger publishers (who should know better) actually embed the subtitles on the video instead of using YouTube's subtitle feature.

Like, if you add the subtitles properly - then it keeps the source video clean, the #subtitles can be community translated, or read by a web browser as text so they can be machine translated or copied and pasted.

Embedding subtitles sucks. And if you are a content creator, plz stop.

I watched the Mario movie with the kids tonight. The story was the same story as 90% of what the kids usually watch, but a bit more developed and with a higher production value (because it's a Hollywood movie and not a Netflix cartoon produced cheaply by starving slave artists.)

However, I loved how the movie perfectly and seamlessly included most of the Mario games and characters that are around.

The best part remained my son reacting to everything. He's always fully invested when watching a story he likes, both emotionally and even physically (jumping around when he gets excited and all). And Mario is one of his favorite characters...

Another interesting thing: the movie (on Prime) only had an English or Japanese track. The kids were surprisingly okay with watching the movie in English with Japanese subtitles. They told me that they're fine with subtitles.
The thing is that at home, the rule is that if whatever they want to watch has a French track, they must watch it in French. However lately, they realized that it'll be easier to watch with Japanese subtitles. I have no objection, it can only help them improve their vocabulary in both languages.
But now, it also means that they're used to subtitles. I didn't expect them to be okay with a full movie in English, but it's great in many ways.

1. It will help them learn English faster.
2. There are so many movies I want to show them when they're a bit older (or some even now, but they're still not interested in live-action movies) and I couldn't imagine showing them dubbed. It seems like I won't have to. Nice.

#movies #parenting #languagelearning #subtitles #Mario

I find it outrageous and incredibly offensive how often subtitles are censored.

I've experienced it on broadcast TV. On Amazon. On YouTube. And elsewhere.

Deaf people do not need swear words hidden. Those of us with hearing loss don't suddenly decide, "Ah, yes, because I can't hear fully, I don't want to experience swearing anymore."

Stop censoring our content!

Wrote two Bash scripts connecting Flite text-to-speech with SRT subtitles, also using bc, sndfile-info, ffmpeg. Probably sndfile-info could be replaced by ffprobe to reduce dependencies.

The first takes an SRT file, and converts it to WAV, adapting the voice speed to match the duration of each subtitle (doing it by stupid trial and improvement, with repeated rendering instead of any fancy calculations).

The second takes a text file with paragraphs separated by blank lines, and converts it to WAV at the natural voice speed, while also putting the actual time stamps with each paragraph into an SRT output file. By default I add a 1/2 second pause between each paragraph.

I wrote two because I found the first one's speed changes too distracting (especially if there are any <tags> in the text, which are not handled properly).

I put the second one here:
https://mathr.co.uk/web/subscript.html

I might clean up and upload the first one too at some point, as it might be useful in different circumstances (e.g. watching foreign films), but my current use-case is authoring video from a text script (so I can change the video to match the computed subtitle timings).

Okay. My search skills seem to be failing me.

I'm looking for a free tool that will:

• Accept the input of a raw text transcript.
• Synchronise that text against the audio file from a video.
• Allow me to export it as a subtitle file.

I've found tools that can do some of the tasks, or do similar activities, but not the exact ones I'm after.

The biggest surprise here is that I'm still considered a young person!

But seriously, sound editors, all of you suck.

The Majority of Young People Are Using Subtitles When They Watch TV - YPulse
https://www.ypulse.com/article/2022/11/22/the-majority-of-young-people-are-using-subtitles-when-they-watch-tv/ #tvshows #streamingtv #subtitles

https://w3.org/TR/imsc-hrm/

Do you need to render complex #subtitles and #captions along a #video on possibly constrained devices such a #TV sets? #timetoimplement

The IMSC-HRM spec defines a complexity model and a quality check for authors to validate that IMSC captions will render smoothly on targeted devices, and for implementers to ensure that their playback system can render the captions at the appropriate time.

We invite feedback! https://github.com/w3c/imsc-hrm/

Looking for #transcription work, so if anyone needs #podcast transcripts, #subtitles for their #YouTube videos, or other random audio you want converted to text, hit me up in DMs or at kimchiteagames@gmail.com. Rates start at .80 USD per audio minute for straightforward files (clear audio, 1-2 speakers, no timestamps) up to $1.50/minute based on complexity.

Super interesting read about character widths and how it affects subtitles.

SRT 32 character line limit doesn't take into account wide/fullwidth characters · Issue #370 · Podcastindex-org/podcast-namespace https://github.com/Podcastindex-org/podcast-namespace/issues/370#issuecomment-1499143266

> Characters in Chinese, Japanese and Korean are twice the width of characters in the Latin alphabet. As such, applying the same 32 character line limit in SRT files to Chinese as you would do t…

New article from Arabic for Nerds:

"The new tool that can even transcribe Gaddafi’s speeches in Arabic"

Maqsam is a new tool to transcribe Arabic videos. It recognizes almost all Arabic dialects. Does it also understand Gaddafi's Zanga Zanga speech? A song by Elissa? A critical test.

Read more: https://arabic-for-nerds.com/tools/transcribe-arabic-videos/

As @fosdem 2023 recorded videos are starting to appear & don't have subtitles, I spent a few minutes of today's lunch break to download a few talks I found interesting & a few I missed, then ran Whisper* on the videos to generate subtitles.

I've shared it at https://github.com/garrett/FOSDEM-2023-subtitles

Feel free to send pull requests with additional talks or fixes.

* Whisper is at https://github.com/openai/whisper — it works on CPUs & is faster on GPUs

For everyone also following #FOSDEM streams this weekend:

There's an app called "LiveCaptions" for #Linux that does realtime subtitles using your laptop/desktop CPU (that is: local, without using the network) for all audio. It'll work for FOSDEM videos too.

It's on Flathub. And it's amazing.

https://flathub.org/apps/details/net.sapples.LiveCaptions