We don’t.

Here in Germany one of the few good things that came out of our previous federal government was the SBGG, which - among other things - allows you to delete the entry. And before anyone points it out: I’m aware it isn’t perfect.

I’ve corrected my previous comments w.r.t. the terminology, thank you.

That being said, assuming those two spectrograms are using a comparable frequency scale, they show quite a bit of a difference: The upper spectrogram shows a near complete “black” stop across all but the lowest frequencies, while the lower spectrogram does not: The 2nd-lowest frequencies continue virtually unchanged, the mid frequencies are only reduced to a dark purple, not completely dark, and only the upper frequencies are truly dark. Those are not the same sound.

Yes, there’s a stop of exhalation between the s and the K. If there wasn’t, it would sound more like the sk of the English word sky instead.

EDIT: He does not, however, stop between Kindern and werden and not between werden and Leute, which does sound weird to me. Less like standard German and more like a dialect.

I would be interested to know still if you hear the glottal stops you’re talking about in the example above, though?

In the voice changed output for “Das Kind”? It’s not perfect, but yes. If you’re interested in the “biomechanics”: It’s the difference between a miniscule stopping of the continuous exhalation from your mouth for the s before starting a new burst exhalation for the k vs doing both as part of the same single exhalation that directly changes from continuous to burst.

The entire new testament is fan fiction. Iēsous of Nazareth didn’t write any of it, the first assumed author never even met him.

The “biblical canon” of today seems to be more or less just whatever was later dictated when Christianity became the Roman Empire’s state religion. Who knows how much other, potentially more accurate reports about Iēsous were lost during that dictatorship?

Now, I agree that Paradise Lost does not have any claim to veracity. I would just also say any claim the new testament has is null and void.

If you provide storage to upload four audio .wav files (each about 400 KiB), I can send you samples generated by me speaking, then cloned with w-okada using a generic voice I believe is freely usable. It’s far from perfect, but it’s good enough for me to clearly tell which is which just by listening.

“Das Kind” is /daskınt/ not /dasʔkınt/.

It absolutely is /dasʔkınt/ in the standard German I’ve grown up with. I’ve never heard anyone any native speaker pronounce it as /daskınt/.

In German, just the same: there is not generally a pause between words written separately: there’s no gap that you can find on a spectrogram of a recording of someone saying, say, “brauner Sack” between the two words.

At the cost of appealing to authority: As a native German speaker, I can absolutely hear the glottal stop a stop between two separate words. It exists. It can also often be found between consecutive vowels within a word, but not necessarily between consecutive consonants like in your example. There’s a reason why it’s in the IPA. It’s also e.g. the basis for the “gender not-really-neutral” dialect of German, which injects an additional glottal stop into a word (like Bürger:innen, which is pronounced like Bürger innen, not like Bürgerinnen).

Mundhirnverbindungslatenz.

Mostly true, but in standard German one would normally vocalize “Schlafsack” without any audible pause between the f and the s.

If you were to pronounce “Schlaf sack”, there would generally be an expectation of a glottal stop an audible stop between the f and the s (dialects excluded).

So while, yes, compound words in German are written without spaces, that’s not just a writing difference.

Be more interesting than a documentary?

The good Apertus model is too large at 70B, unfortunately. The 8B model is much too small to be useful for what I need.

Yeah, it seems the llama.cpp changes have been merged in the last 24 hours. I’m hoping ollama makes them available this week.

Merzweckeier!

Thank you very much, I hadn’t heard about this model coming out anywhere!

Seems like I’ll need to wait just a bit more for it to be available in Ollama. As soon as it’s there I’ll try it out :)

In the USA (which was the focus of OP) I would argue there is virtually no left in the first place. Outliers excluded even their left-most people are still centrists, at best. Note that nowhere I said people should stop voting, it’s kind of crazy how that’s always a knee-jerk assumption. Obviously you still ought to choose the lesser evil - but that must come with doing the other parts. Otherwise, again, you’re just choosing evil.

That being said: If you live in the USA and you want that to happen, you need to step up. Not say “nobody else is doing it”. You need to do it. If you already are: Congrats! Next you need to accept that you can do everything right and still lose. That’s just life.

I’m really hoping Soofi models are going to be good. Because afaict they’ll be the first (and so far only) open source (not just open weight) models in the ~30B parameters range.

My point is the nay-sayers here say “don’t vote”, but offer no alternatives.

I’m a nay-sayer (to doing no more than perpetually voting for the lesser of two evils), and I’ve offered alternatives. I’m also not alone in this, I’ve seen about as many people, proportionally speaking, advocate for the same things in past posts about this here on Lemmy as there are on Reddit. Might this be a bubble issue on your end?

You say join a union? I say hell yeah, sign me up! Can I join yours?

Feel free to come to Germany and join one of our unions. The one primarily responsible for negotiating the collective agreement currently determining my salary is ver.di.

False dichotomy. There are many things to do between “do nothing but vote for evil” and “revolution”.

I don’t want to write the whole thing again, so I’ll just leave this here.

To add: I’ve never tasted an alcoholic drink that didn’t improve by removing the alcohol.