Well, I guess it's no surprise- the poor quality of audio on these vids sometimes, the endless variations of dialect. Slang, background noise, what have you. Still, it can be amusing at times.

Here's an example. A Wal-Mart manager is offscreen saying "I'm concerned about the safety of my customers" but that's not how the software interpreted it:





I did a bit of research and I think I see part of the problem. Here's the author of the CC auto-generator program:

I saw a thread on a forum talking about CC on tv and remembered this picture.

I was watching a car repair video on youtube early one morning without sound since my wife was sleeping next to me. I about jumped out of bed when the close captioning read my brother's first and last name for what should have been "pad wire"

During a fight scene with lots of fight action, but no other sounds.
Auto-generated captioning: "(music)"
Not even a "Biff!" or "Pow!"

It's a little more understandable with Youtube, due to the sheer volume of new videos being uploaded every minute. There's no way they could keep up with manually adding captions.

But Amazon Prime Video has the same issue on some of their content. I guess they figure the cost savings of having a computer do it and only being 80% accurate is worth it, compared to paying a team to actually generate 100% accurate captions.

I mean, are not we all concerned about the safeties of our assholes?

This is outstanding, and it’s very similar to the voicemail to text feature on my iPhone. Can’t blame the translation though because one can’t be too careful these days.