You've probably never watched Indiana Jones speak French... I was forced to when staying for 2 weeks in the south of Belgium as part of a French immersion program. It's unbearable.

When I watch an American movie, I want to hear it the way the director intended it to be. I don't want every villain in every movie have the same voice. If I want to hear Dutch in a movie, I watch a Dutch movie. It's not that deep.

The fact that it helps kids learn a different language is a very nice fringe benefit.

I remember watching an English movie with an incorrect subtitle in school when I was 12, well before my first English class. The whole auditorium laughed because everyone caught the error.

> I don't want every villain in every movie have the same voice.

Sounds like your problem is with crappy, cheap dubbing, not dubbing in general.

Look at Disney animation for dubbing done right (and more so in the 20th century - these days I’m not so sure).

Yeah, because I totally have control over who’s dubbing movies in France.

A cartoon is the worst possible example.

Tell us more.

I don't think that generalize to dubs as concept itself, I know for the fact that Japanese dub of Commando(1985)[1] is quite something.

1: https://en.wikipedia.org/wiki/Commando_(1985_film)

The thing is so unreadable one would think that it's a parody of an LLM article.

I'm curious about what actually happened, but I can'make it past the second section, the writing is so terrible.

My theory is that YouTube blocks some accounts for publishing LLM-generated music, and people who wanted to earn ad money from it get burned and publish LLM-generated posts about it.

I would be on YouTube's side here, except it's possible that their motivation is simply to avoid poisoning their dataset while they train their models off creators videos. Also, the question is how they tell apart what's LLM-generated without false positives.

Maybe there were also artificial listens fraud (it's a problem with their competitor Spotify), but we'll never know because no one who was blocked would publish that honestly.

Is buying hopefully broken electronic test equipment at the flea market, fixing it and then blogging about it a niche hobby?

of course

Minor usability comment: the screenshots are too small to be readable. Whenever that's that case in my blog posts, I make those screenshots clickable and add (Click to enlarge) below it, to make it easier for readers to see the image are original resolution. In markdown, I do that like this:

[ ](image_url.png)

(Of course, I can also right-click and do "Open image in new tab", but that's one click extra...)

Congrats on the awesome project, BTW! You were lucky that I wasn't sitting next to you on the plane. I would have wasted so much of your time asking dumb questions.

Images now expand when clicked - thanks for the feedback!

As an experiment, I just now took a random section of a few hundreds bytes (as a hexdump) from the /bin/ls executable and pasted them into ChatGPT.

I don't know if it's correct, but it speculated that it's part of a command line processor: https://chatgpt.com/share/69d19e4f-ff2c-83e8-bc55-3f7f5207c3...

Now imagine how much more it could have derived if I had given it the full executable, with all the strings, pointers to those strings and whatnot.

I've done some minor reverse engineering of old test equipment binaries in the past and LLMs are incredible at figuring out what the code is doing, way better than the regular way of Ghidra to decompile code.

I've cut-and-pasted some assembly code into the free version of ChatGPT to reverse engineer some old binaries and its ability to find meaning was just scary.

Yesterday, i had claude decompile and fix firmware for my new samsung viewfinity s8 - there was really annoying pop up banner on each wake which you cant turn off, and samsung clearly didnt care. I was about to return it, then thought - hhmm, why not :) Not one-shotted, took several tries (lucky none of them bricked it, haha). Also i guess warranty is voided, but idc :)

Their stash will explode. LLMs can do this on binaries just the same, and there's a lot more closed than open source SW out there.

And they also have a nearly infinite budget to rent AI time to do this type of work.

> Once the design gets past toy size,

Do you consider 800+mm2 slabs of 3nm of silicon still toy size? Because there's a very high chance that those were written in Verilog, and I've never had to chase sim vs synthesis mismatches.

> Verilog gives you enough rope.

Yes. If you don't know what you're doing and don't follow the industry standard practises.

> Why don't VHDL and Verilog just simulate what hardware does?

Real hardware has hold violations. If you get your delta cycles wrong, that's exactly what you get in VHDL...

They're both modeling languages. They can model high-level RTL or gate-level and they can behave very different if you're not careful. "just simulation what the hardware does" is itself an ambiguous statement. Sometimes you want one model, sometimes the other.

AFAIK, creating latches is just as easy in Verilog as in VHDL. They use the same model to determine when to create one.

But with a solid design flow (which should include linting tools like Spyglass for both VHDL and Verilog), it’s not a major concern.

SystemVerilog basically fixes this with always_comb vs always_latch.

There's no major implementation which doesn't handle warning or even failing the flow on accidental latch logic inside an always_comb.