• NephewAlphaBravo [he/him]@hexbear.net
    link
    fedilink
    English
    arrow-up
    61
    ·
    edit-2
    7 months ago

    if AI ever gets good enough to do this accurately and in real time, we’d be looking at an actual babelfish

    i would finally, at long last, have to hand it to them

    • GarbageShoot [he/him]@hexbear.net
      link
      fedilink
      English
      arrow-up
      56
      ·
      7 months ago

      Sentence structure means that it kind of can’t happen in real-time as such, because you would need to wait until potentially the end of the sentence to get words that appear early in the sentence in an accurate and natural-ish translation. If “20 seconds later” is real time, barring run-on sentences, which are much more common in speech than in writing, then I guess.

    • Kaplya@hexbear.net
      link
      fedilink
      English
      arrow-up
      35
      ·
      edit-2
      7 months ago

      Yandex Browser already does this, but to Russian only. It has like 10-15 seconds delay for live streams (at least on Youtube) but it works as well as the auto-generated transcription.

  • Kaplya@hexbear.net
    link
    fedilink
    English
    arrow-up
    57
    ·
    edit-2
    7 months ago

    Here’s the funny part: their American accent totally made it believable.

    It’s very clear that even with the AI generated voice, they are not native Mandarin speakers. They sound like your typical foreigners who learned Chinese for a number of years lol. I don’t know if it’s the dataset they’re trained on or just how the algorithm works, but it’s very interesting.

    • WayeeCool [comrade/them]@hexbear.net
      link
      fedilink
      English
      arrow-up
      35
      ·
      edit-2
      7 months ago

      Makes me think about what it would be like if Chinese ever becomes an international language, in the way English has and Latin did before it. It makes me giggle to think about Mandarin with a backwoods Tennessee drawl.

    • gobble_ghoul [he/him]@hexbear.net
      link
      fedilink
      English
      arrow-up
      8
      ·
      7 months ago

      Even with the phonemes of any two given language varieties that are considered to be “the same sound”, there are going to be differences in what the average pronunciation is, so I assume that’s a lot of what’s going on here. The other thing is that English and Chinese have a lot of phonemes that barely or don’t at all overlap in possible pronunciations, so the algorithm is picking the closest match.

  • grandepequeno [he/him]@hexbear.net
    link
    fedilink
    English
    arrow-up
    52
    ·
    7 months ago

    Felix in the replies: “I’m crying at how beautiful this is. I support AI now. all I have ever wanted is for the show to be credibly portrayed as a Chinese podcast”

          • marxisthayaca [he/him,they/them]@hexbear.net
            link
            fedilink
            English
            arrow-up
            32
            ·
            7 months ago

            don’t be too hard on yourself - a) he probably didn’t read all of these and it is some intern’s job to make this list b) reading for fun and reading for information are too different skills. You can buy a book on philosophy, skim 50% of it and deep dive into a single chapter.

          • Fishroot [none/use name]@hexbear.net
            link
            fedilink
            English
            arrow-up
            4
            ·
            7 months ago

            Various people who knew Andropov well, including Vladimir Medvedev, Aleksandr Chuchyalin, Vladimir Kryuchkov[92] and Roy Medvedev, remembered him for his politeness, calmness, unselfishness, patience, intelligence and exceptionally sharp memory.[93] According to Chuchyalin, while working at the Kremlin, Andropov would read about 600 pages a day and remember everything he read.[94] Andropov read English literature and could communicate in Finnish, English and German.[95]

        • Judge_Juche [she/her]@hexbear.net
          link
          fedilink
          English
          arrow-up
          8
          ·
          7 months ago

          On the one hand it’s very lib to be reading Piketty; but on the other Ninety-Three is Hugo’s best and most revolutionary novel and dosen’t get a lot of attention becuase of it. So I guess it’s a wash.

          • Fishroot [none/use name]@hexbear.net
            link
            fedilink
            English
            arrow-up
            5
            ·
            edit-2
            7 months ago

            Piketty is a one trick pony, I went to one of his conference at my University. All he advocates for is a global progressive tax system. He never explains how it is going to be implemented (he himself acknowledges that there is a need for all government to form a UN like IRS which he doesn’t believe it’s ever going to be possible) or the fact that the implementation of his global tax system would just nuke most of the economy of countries that are just taxe havens for the rich.

            The Netflix documentary of the book is enough to explain its contents. It’s funny to see him talk about capitalism without mentioning ONCE Marx, Webber, Ricardo, etc. Or why the USA did the New Deal (because they never mention the Bolsheviks Revolution at all). The only mention of the USSR is that ‘‘wow it collapsed because it’s a failure and a police state, oh wow suddenly wealth inequality skyrocketed for no reasons’’

            Oh yeah, China is basically doing State Capitalism not so different than USA Post WW1 (never going into the details ofc), but they are bad because there are wealth inequality

    • Kaplya@hexbear.net
      link
      fedilink
      English
      arrow-up
      31
      ·
      7 months ago

      Native Mandarin speaker here. They all sound like your typical Westerners who have lived in China for a number of years. It’s more interesting that the AI were able to give them that realistic Western accent than a proper regional Chinese accent.

        • Kaplya@hexbear.net
          link
          fedilink
          English
          arrow-up
          16
          ·
          edit-2
          7 months ago

          Accurate with regards to which one?

          Even without watching the clip, any native speaker will be able to immediately tell that they’re not native Mandarin speakers, as the sounds do not correspond to any regional accents in China or even Taiwanese and other Chinese diaspora accent (Malaysian, Singaporean, for example).

          As others have said, there are some subtle finesse in the AI voice generation that are very interesting - like Will’s “yi dian er” (一点儿), but they still sound like foreigners who try to imitate Chinese speakers.

          In other words, if the Chapo boys move to China and live there for 5-10 years, they’d probably sound like that. I am more impressed by how “realistic” the AI voice imitates Westerners speaking Mandarin as a foreign language than just transposing perfect native accent on to the input English sentences. I don’t know how the algorithm works but it’s very interesting.

          Will sounded like he has lived in China for 5+ years, Matt’s accent is slightly worse, so probably 2-3 years. Felix sounded like he just arrived last week lol (just extrapolating from my own experience with foreigners, as we know everyone learns at a different rate).

          • Tachanka [comrade/them]@hexbear.net
            link
            fedilink
            English
            arrow-up
            4
            ·
            edit-2
            7 months ago

            I meant accurate not regarding their dialects but regarding the translation itself. As in was it accurately translating what they said? Sorry, I should have been more specific.

        • CloutAtlas [he/him]@hexbear.net
          link
          fedilink
          English
          arrow-up
          14
          ·
          7 months ago

          Getting the “er” in there is quite northeastern, my grandfather speaks with 儿化. Adding a soft “er” sound at the end of syllables, basically.

          But they also don’t have 100% correct tonal pronunciation, something like 80-90% correct.

          This is actually extremely impressive.

          • Fishroot [none/use name]@hexbear.net
            link
            fedilink
            English
            arrow-up
            3
            ·
            7 months ago

            The ‘‘er’’ is not profound enough to be authentic, it is the exact mistake that Dongbei people would pick up on people who are from the south for example

    • Fishroot [none/use name]@hexbear.net
      link
      fedilink
      English
      arrow-up
      2
      ·
      7 months ago

      Will actually sounds like a mango that picked up some beijing accents because he worked at a dongbei style restaurant for 1 year.

      Matt sounds like someone from Guangxi

  • silent_water [she/her]@hexbear.net
    link
    fedilink
    English
    arrow-up
    23
    ·
    edit-2
    7 months ago

    how good are the translations? does anyone know Chinese and can compare the original to the translated version? I’d be interested to hear how accurate the translations are. passingly accurate transcription services are a huge boon.

  • Tachanka [comrade/them]@hexbear.net
    link
    fedilink
    English
    arrow-up
    21
    ·
    edit-2
    7 months ago

    If we get to the point where this kind of stuff can process accurately, in real time, on a mobile device, we will have destroyed the need for real time speech translators (sorry, translators), and, more importantly, most of the language barriers between the international working class, no? marx-goth Cautiously optimistic.

    • Kaplya@hexbear.net
      link
      fedilink
      English
      arrow-up
      19
      ·
      edit-2
      7 months ago

      I already wrote this below, but Yandex Browser already does this! It only translates to Russian, and with live streams (on Youtube for example) you get a ~15 seconds delay.

      It’s basically as a real-time transcription -> translation -> voice generation pipeline so the accuracy is as good as the transcript it manages to extract from.

    • Timberknave [none/use name]@hexbear.net
      link
      fedilink
      English
      arrow-up
      17
      ·
      7 months ago

      I am not that scared for my job in the next 10 years. As long as people don’t trust Bazinga Translate from the Torment Nexus Company to translate the doctor who goes through the Operation details with them, I am safe. Remember many countries want a stamp from a sworn translator for legal documents. Not to mention how shit even AI translation is still for Arabic (if it can recognize the letters to begin with, lmao). I can pre-translate a text and then go through it again, sometimes it even saves me time.

      • SerLava [he/him]@hexbear.net
        link
        fedilink
        English
        arrow-up
        12
        ·
        7 months ago

        Yes I think this technology will be great for casually listening to shit, but won’t get in the way of serious translation that matters.

    • space_comrade [he/him]@hexbear.net
      link
      fedilink
      English
      arrow-up
      5
      ·
      7 months ago

      Eh I dunno speech to text algorithms are still kinda garbage. They work well on simple sentences but once you throw in colloquialisms and abbreviations it just craps out.