kbin.social has been totally down for a while. I don’t think your posts are actually federating when you post into a kbin.social magazine right now; the votes you are getting are probably from other lemmy.world users only.

I don’t know how to do it with KDE’s tools, but on the command line with ffmpeg you can do something like this:

ffmpeg -i video_track.mp4 -i audio_jp.m4a -i audio_en.m4a -map 0:v -map 1:a -map 2:a -metadata:s:a:0 language=jpn -metadata:s:a:1 language=eng -c:v copy -c:a copy output.mp4

Breaking it down, it:

• runs ffmpeg
• with three inputs (-i flag) – a video file, and two audio files.
• The streams are explicitly mapped into the result, counting the inputs from 0 – i.e. -map 0:v maps input 0 (the first file) as video (v) to the output file and -map 1:a maps the next input as audio (a), etc.
• It sets the metadata for the audio tracks -metadata:s:a:0 language=jpn sets the first audio track (again counting from 0…) to Japanese; the second metadata option sets the next audio track to English.
• -c:v copy specifies that the video codec should be copied directly (i.e. don’t re-encode – remove this if you DO need to re-encode)
• -c:a copy specifies that the audio codec should be copied directly (i.e. don’t re-encode – remove this if you DO need to re-encode)
• output.mp4 – finally, list the name of the file you want the result written into.

See documentation here: https://ffmpeg.org/ffmpeg.html

If you need another language in the future, I think the language abbreviations are the three letter codes from here: https://en.wikipedia.org/wiki/List_of_ISO_639-2_codes – but I’m not certain on that.

Now imagine if to buy a car you had to tolerate cameras and other forms of tracking your telemetry just to get to work and feed yourself.

Sorry to be the bearer of depressing news, but that’s basically already happening in new cars.

https://foundation.mozilla.org/en/blog/privacy-nightmare-on-wheels-every-car-brand-reviewed-by-mozilla-including-ford-volkswagen-and-toyota-flunks-privacy-test/

https://jacobin.com/2024/03/car-spying-insurance-surveillance-data/

A fairly vocal portion of lemmy is AI-hostile, and even for the people who aren’t outright hostile to it, it can be annoying at times – AI content does tend to drown everything else out when it’s permitted, so making a community explicitly for it would probably work better.

lemmy.dbzer0.com might be a good place to host a community specifically for exploring AI generated music if you’re interested in running one. That instance is explicitly open to AI gen and already has several image gen communities, but I don’t think they have a music gen community yet. (Double check though before making one in case I just missed it.)

I wonder what cuil things it will say if you start asking questions about hamburgers instead…

I ran into an example of the thumbnail issue again today – this time on a post from kbin: https://old.reddthat.com/post/19193476

The thumbnail looks like this in the HTML:

<div class="thumb">
  <a class="url"
     href="https://media.kbin.social/media/60/a4/60a45b8ff88b1b2e3a0f77b701feb323c5bbfb7ceeb75154ea7df5d6eea15ef8.jpg"
     >
    <div  style="background-image: url(https://media.kbin.social/media/60/a4/60a45b8ff88b1b2e3a0f77b701feb323c5bbfb7ceeb75154ea7df5d6eea15ef8.jpg?format=jpg&amp;thumbnail=96)"></div>
  </a>
</div>

Note that it’s making a request to kbin.social with ?format=jpg&thumbnail=96 parameters in the CSS – which results in the full image being loaded since kbin doesn’t run pictrs.

The versions in use on reddthat (according to the settings page) are:

lemmy: 0.19.4-beta.7

mlmym: 0.0.44

That requires turning every read into a write – which is slow/expensive generally. (That might not matter much for Google – who try to record everything you ever do already, basically – but it matters for everyone else.)

Also, it tends to promote spam and offensive niche content. kbin’s got a sidebar that tries to promote random low activity communities and posts, for example, and it’s almost uncanny how much crap it pushes up…

Mrs Bighead?!

…

*hangs up the phone*

Personally, I prefer it when people do one of the following:

• upload the file to catbox.moe and link it here (for clips up to 200MB)
• upload the file directly to their lemmy instance (if their instance allows it)
• self-host it on their own web server (with no bullshit crappy JS interface, please – just give me the file; I’ll play it with VLC if it doesn’t work in my browser)

PeerTube is also a reasonable choice – although I don’t like its UI very much.

I’m having trouble finding an example of the thumbnail issue again right now but I was seeing the pictrs conversion parameters passed to URLs from catbox.moe, i.postimg.cc, and other sources in the CSS for the thumbnail when I reported the issue to Tiff ~3 weeks ago. It’s possible that it got fixed/suppressed by another change since then though. (0.0.44 was deployed a few hours ago and I think there may have also been a beta patch bump for the lemmy backend at some point since I reported the issue originally in our local support community.)

I’ll let you know if I see it pop up again.

For the text handling issue, I was seeing text like “<thread title> by <username> in <community>” (i.e. "<thread title> by <username> in <community>" if it still happens) getting misinterpreted as raw HTML instead of being escaped. (i.e. <!-- raw HTML omitted --> was showing up in the HTML output for the page.)

You may recognize that text as the pattern for a recently fixed bug in the user profiles; I found the text handling issue while trying to explain the other issue to Tiff a few weeks ago.

Will edit this comment immediately after posting to let you know if I still see the text issue.

EDIT: I still see the text issue show up in this comment. https://old.reddthat.com/comment/10370610

Thanks! I’ll go ask Tiff about getting reddthat updated later.

BTW, is there a community for discussion of mlmym itself somewhere on lemmy? I can’t participate on GitHub, but those aren’t the only issues I’ve found. (e.g. there’s also ?format=jpg&thumbnail=96 on non-pictrs links and a text handling issue with angle brackets…)

That’s great to hear!

You can open any profile with multiple pages worth of posts or comments on old.reddthat.com and it’s jumbled. Even my own profile is jumbled: https://old.reddthat.com/u/e0qdk

The first page is mostly comments I made two weeks ago plus a thread from today and some very old threads. The second page has comments I made earlier today and during the past week. The third page starts with my most recent comment and then has a bunch of older comments.

The exact order might change after posting this, but my own recent comments mostly being on page two has been pretty consistent for a while.

If I look at a very active user’s profile (like MentalEdge’s), I see threads from today show up on page three(!) while there are threads from a week or more ago on pages one and two.

I’m not sure what’s going on exactly, but it basically makes user profiles pretty useless right now through mlmym.

Edit: I can’t even find this comment in my profile, but my other reply (regarding the envelope being fixed in 0.0.43) shows up on page 3.

I haven’t had much issue with lag, generally, but I don’t get notifications any more – which is probably the most pressing issue. (I have to remember to manually check once in a while after I post since the envelope doesn’t light up.) That might be an issue with reddthat being on a recent beta version of lemmy – I don’t know.

We do have lemmy-ui-next over here too. Thanks for reminding me about that. I’ve been meaning to poke at it a bit.

Right now I’m mostly using mlmym (the “old” interface on most instances that support it) because it doesn’t require JS for basic viewing.

It’s kind of buggy though, unfortunately – things like user history show up as a complete jumble, for example. :(

One of these days, I’ll probably get fed up enough to go write my own interface and set things up exactly how I want them to work… but I’ve got too many projects already so I’m just living with it for now.

I wonder if this will actually cause an increase in the number of security vulnerabilities and breaches as there’s now a fairly obvious way for employees to penalize their bosses financially for being assholes…

My old username from reddit and HN was already taken and I couldn’t think of anything else I wanted to be called so I just picked some random characters like this:

>>> import random
>>> ''.join([random.choice("abcdefghijklmnopqrstuvwxyz0123456789") for x in range(5)])
'e0qdk'

I have that literally in my kbin profile, but it’s not on my reddthat one. (I think I tried to copy it there originally when I set up the account but ran into some issue with Lemmy’s UI – been long enough that I forget what exactly.)

I think this is just using SpeechDispatcher from the system – so it’s not a Firefox specific thing. I get a similar (but very slightly different) voice on my own system by default – which matches what I get when I run a command like spd-say --wait "Hello world" from the command line.

I’m pretty sure SpeechDispatcher can be configured to use a different synthesis engine – Arch’s wiki has some suggestions: https://wiki.archlinux.org/title/Speech_dispatcher – but I haven’t dug into it yet.

Maybe you’d be interested in “kinetic novels”? They’re basically VNs without choices.