Nimitz14

Or Diabotical. Free on epic games store.

edit: nevermind

It's blindingly obvious that you would only ask this to someone who should know it.

Thanks for mentioning this, because after reading the answer my first thought was "wait, but what if I do nothing before doing IFFT which is also equivalent to multiplying by a rectangle (but that nonzero from -nyquist to nyquist)"

Fighter pilots seem quite positive about it.

Then pay for it.

I checked about of half of these out and I can't take any of the complaints seriously. The strongest argument against tracking remains the moral one.

Telecom companies could have been the best speech recognition companies out there. Instead they did nothing until recently. Your perspective lacks imagination.

And you were likely much less efficient because of less interaction with your more knowledgeable colleagues. You don't know what you don't know.

No, a teenager is not a child.

Common sense seems to be a rare quality these days. Obviously if it's necessary to do the job then you talk about politics.

Thanks for sharing. I had no idea stuff like C&C, Total War and CoH are on mobile now. That's insane.

If you went back twenty years one could also compile a similar list (not necessarily remakes, but something of a similarly low quality). Pretty clear you just have an axe to grind.

Some years are better than others. Of course 2021 was going to be shitty, have you been living under a rock and not heard of covid?

2019 was a great year for movies for example.

This is the most whiny and ridiculous post I've ever read on hn.

Ridiculous statement. The homocide rate is so much higher in the US. Hard to believe anything you say.

one word: pybind11

I honestly think people are severely underestimating what a massive impact this is currently having in increasing productivity of python devs who know a little C++.

Yeah... I'm rather perplexed why the boss didn't just fire him on the spot. A new grad is likely not valuable enough to be worth fighting for.

Could you elaborate on this? I thought Aluminium is relatively easy to manipulate (and that's why it's used a lot)? Why wouldn't one use injection molding?

Sorry but describing Umberto Eco as a pop writer is simply hilarious.

If you have the data and a specific domain you can focus on then building a custom model [with kaldi] should always win. That's what I've done in the past (beating google, nuance etc.). You most likely didn't have the data and/or didn't know kaldi well.

> Obviously Nuance is more than just speech recognition, but still not sure why people are downplaying how good they were at it.

Because nuance wasn't very good.. at least in all the benchmarks I've seen. It's been a while since I compared numbers it's possible they've improved a lot. They're also known for kinda being dicks with the contracts they offer in B2B.

That answer depends on the language and on your use case. It seems like you're asking about desktop apps, but my parent was not talking in that context. Indeed there's not a lot of choice there because there's no money in it.

I don't think this is accurate (and I work in this field).

MS has close to zero presence with regards to papers and few industry positions. Them building a (bad) toy system means little. Google and FB are on a totally different level in comparison.

I have a vague memory of reading that the higher ups just didn't believe in voice-based interfaces.

Well that's a surprise. MS used to not really be interested in speech, guess they changed their minds?

You probably aren't aware in Europe flights can be like 20E. The cheapest flight I ever had was London to Madrid for 3 pounds.

Oh wow, I've definitely done this mistake without realizing it...

Since we're trading anecdotes: I'm European and I disagree.

> In the jobs that I do, "bad performance" is super subjective and very hard to measure.

Then either you and all your coworkers so far have been amazing, or you're the bad one. Either way consider yourself lucky!

> And yet that is not what is observed in practice. See figures 1 and 2 of that paper

Yeah... that's why the paper is a good contribution contrary to what you're saying. Not sure why you're repeating this information.

> What I am complaining about is that the authors are confusing "what can be represented" by a neural network with "what can be easily learned" via SGD. I argued that the peak in test loss in those figures is observed becuse SGD struggles to find a solution that generalizes well, not because those networks are intrinsically less powerful (as the paper seems to imply).

I mean.. those models are by definition less powerful as they have less parameters. The (to me) main point of the paper is to point out a symptom which is interesting. Their explanation for the symptom being (maybe) wrong doesn't detract from the important work of showing the symptom exists.

A legitimate criticism would be that there have been earlier papers showing the same thing.

Sidenote: It seems having a larger model does make it easier for SGD to find good solutions[1]

[1] https://www.youtube.com/watch?v=kcVWAKf7UAg