Forgot your password?
typodupeerror

Comment Yeah, automatic subtitles are pretty bad (Score 1) 100

I did some video where I had a script. So it was easy to extract subtitle and compare to script. Yeah, its pretty bad. Probably every other sentence had some obviously incorrect subtitling. And these videos were on clean audio inputs: no background noise, good mic.
I ended up having to correct the subtitles using the transcript.

I have seen a study recently on noisy audio in a group setting, so the speakers maybe not be perfectly mic-ed and all. The speech-to-text was done with a couple of Whisper models. The error rate was something like 30% of the transcription was bogus. (yeah, 3 words in 10.) I imagine that YT would do some preprocessing. But overall, I find the subtitles not terribly good. It's useful and better than nothing.

Submission + - A coalition in the EU is building Euro-Office as an alternative to MS Office

thephydes writes: It will be interesting to see how this progresses.

https://tech.eu/2026/03/27/eur...

"Across Europe, public administrations, enterprises and educational institutions are reassessing their dependence on non-European productivity platforms. While office software remains mission-critical infrastructure, there is currently no solution that combines full Microsoft format compatibility, a familiar user experience and genuine digital sovereignty under European stewardship."

Comment Re:reason why (Score 1) 26

Abso-fucking-lutely!

The JS ecosystem (and frankly the python ecosystem too) is made of lunatics who think I have nothing better to do than patch my system every 3 month because they decided that it was more pretty to swap the order to parameters somewhere.

The number of NPM packages that are deployed widely and that just break at an upgrade is staggering. And because fucking JS, you don't get a proper error message at the upgrade recompilation. You need to rely on having a test that catches the difference. What a fucking nightmare of an ecosystem.

I am currently the maintainer of a small project but we can't break production. We did our major upgrades in summer 2024. So we used the most up to date release, v22, that came out 3 month before. This was 18 month ago, and its EoL is in about a year. Who has the ability to commit the engineering time to upgrade every single one of your services every 3 years?

These are not minor changes, these are going to be different majors. So I do expect some feature break. The last update which was also within 3 years caused a forced upgrade of database driver and abstraction layer to a new major. And they decided to rewrite their entire fucking API.

I do not understand web people. How do they not go insane because of the horrible engineering decisions they keep on making in that field? You know what, the shitty Perl webservice I wrote in 2010, it still fucking works! And no one has touched it since.

Comment Not terribly surprising (Score 1) 45

DYI anything is usually an order of magnitude cheaper than a professional production.

Most movies these days seem to spend as much money promoting the movie than they spent producing the movie. Here promotion was free to him. So that's a factor 2 right there.

DIY amateur does not mean it's bad. In video games, we've seen that story unfold a hundred times in the last 20 years.

Comment Re:I thought this happened a few years ago (Score 2) 3

yes. Heroku has become very expensive. I migrated a few of our services to a native EC2 box for 75% cost reduction within the last year.

I still only have a handful there for a services there where it is a bit harder to migrate because we did our setup stupidly. (We put the heroku domain name in client installations that are harder to upgrade.) But we'll bite the bullet and migrate out this year I would assume.

Comment Re:A lot of them are weird religious schools (Score 2) 146

I teach at $stateuniversity. We have been expecting this for years. For us the number of 18 years old has peaked in 2023 I think. And the enrollment prediction of freshman in the state are within 2% of the prediction we made in 2019 purely based on population.

Ironically, we are getting more students. What is happening is that when the total number of student regionally decreases, we did not lose student. But hyper local small universities lost students to us. We admitted the same fraction of students give or take, but a higher fraction of student came to us rather than a tiny school so our yield went up. Also some local universities had to shut programs down and the students who would have gone there are going to us instead.

The dynamics is interesting.

Now I am not saying that other factors aren't part of the equation. But if you ANOVA it, I bet it's 95% an effect of number of 18 year olds.

Slashdot Top Deals

"I've seen the forgeries I've sent out." -- John F. Haugh II (jfh@rpp386.Dallas.TX.US), about forging net news articles

Working...