Originally published at: https://boingboing.net/2018/02/16/epic-chains-ad-explains-benefi.html
…
1 minute in and I know what the YouTube comments are going to be like.
How 20th century. Couldn’t those boats use digital anchors?
May I recommend the AdBlock Plus and Element Hiding Helper plug-ins? A few clicks and all YouTube comments disappear forever.
It’s way better to watch without knowing how the trolls are abusing the audience.
This guy would’ve cut some badass promos at that factory. RIP
Link is broken
The combination of the guy’s vocal tone and his highly accented pronunciation comes off as really strange. It’s like he’s doing American Announcer Voice but it’s garbled. I would probably do better understanding him if he used a more natural Chinese-accent tone.
As someone who has done some freelance writing I try to avoid using ad blockers.
I wasn’t going to scroll down to the YouTube comments anyway.
anything pocket-watch sized?
I think it’s a computer translation, but I can only wonder what the input was. That doesn’t sound like any natural accent I’ve ever heard.
I think this is the result of using some sort of Asian text-to-speech engine with English text. Note how the Chinese city names are pronounced with seeming accuracy (I say “seeming” because my Chinese language expert isn’t available) but simple words like “popular” are rather creatively mangled.
The text is well written idiomatic business English, so you wonder why they didn’t just have the person who wrote it do the sound track…
My guess is that it’s a text-to-speech engine designed for Chinese. TTS engines are finally to the point that they can generate speech that is very close to sounding natural, but the catch to that is the tuning needed to make a synthesized voice sound natural tends to be fairly language specific.
The timbre of the voice is so good that I’m guessing this was generated with a legitimately good tool–it’s not the timbre of the voice that is unrealistic, it’s the delivery of the words themselves. The amount of accent that an individual has when speaking a foreign language varies from person to person, but those variations tend to fit within a certain set of parameters which are established by aspects of their first language. For example, many native English speakers struggle with things like “R” sounds in Spanish and “ẞ” sounds in German because English doesn’t really use those sounds. Another near-universal problem for people speaking a second language is minimal pair confusion.
The atypical flow, pronunciation, and inflection of the narrator aren’t consistent with a Chinese person speaking English, but they do seem to be consistent with a TTS engine that has had to be fed phonetic pronunciations for most of the words. One of the giveaways is the fact that modern TTS engines tend to have dictionaries to recognize common words that require specific inflections or emphasis. The thing that made me pretty sure was the way the narrator all but skips the word “the.” The pronunciation is very soft and quick, almost like the end of “fifteenth” but with a very short, soft, “e” at the end. Modern English TTS engines wouldn’t make that mistake because English doesn’t pronounce the word quite the same as it looks on paper.
I doubt a TTS engine meant for Chinese would have a dictionary for English words like that.
My guess? It’s substantially easier to become literate in another language than it is to learn how to competently speak it. It seems probable that someone at the company wrote the text and used a sophisticated TTS engine to create the dub. I don’t know what the given reasons were, but from the timbre of the voice, I’m guessing that the TTS engine is actually very good, but just not designed to handle English words. I suspect it would be pretty difficult for a person who is literate, but not fluent in a given language to find all of the oddities in pronunciation and cadence of a recording of that language even if their ability to read and write that language is fairly strong.
I read somewhere that everyone’s using blockchains these days.
As a committed flasher, when someone exposes themselves to my kids at the playground, I insist that they watch respectfully for at least a full minute
No, I’m pretty convinced this wasn’t TTS. It doesn’t have any of the telltale cuts you get in the best examples of machine generated voice. I’m pretty sure this was a Chinese voiceover guy doing his very best American accent possibly with a phonetic translation but just not getting there. It reminds me of those odd DPRK English-language propaganda videos where the guy speaks grammatically perfect English that still sounds super strange due to his pronunciation (in that case, pronouncing Korea Korear, etc).
Needs more cowbell.
Sez you!
Rob has the weirdest fetishes…