We can earn a commission from left on this page.
If you look at a media piece that is produced in a language you don’t speak, you have a few options. Of course you can switch on subtitles and participate by reading. In some cases, however, you can switch the audio track to the language you speak. This is referred to as “dub” and usually requires that language artists record the lines in this new language – in a way that corresponds to the original tone of the show or film.
This requires time and money, something that studios are not always willing to separate. As such, not everything has a dub. But what if it could?
On Wednesday, Amazon announced “Ai-side synchronization” for Prime Video. With selected titles, the audience can play an A-generated DUB of the script in English or Latin American Spanish. Amazon says that there are 12 titles that support this function at this time, but only three called Three: El cid: the legend; My mother Lora; And Long. The company also says that “localization professionals” work with AI to “ensure quality control”. It is not clear what that means unless there are people who check these AI dubs to ensure that they are not transferred incorrectly or make other mistakes Due to hallucinations.
I have first -class video, so I was interested in seeing these dubs for myself to see how Amazon’s AI compared to people compared to people. I couldn’t find my search My mother LoraBut I could find the other two titles. To use the function, you must enter into the compatible title in the voice selection and select the “AI Beta”. Then you can experience what I have experienced.
El cid: the legend
There are a few titles for Prime videos with the name of The cidBut the only one who supports AI synchronization is This specific option: A 63 -minute documentary The cid. I didn’t watch El Cid: Layend To learn more about this ruler from medieval Spain – I wanted to hear how well an AI bot would deal with the English synchronization of a film produced in Spanish.
The results, to put it the least, are interesting. Most of the speaking in the documentary about Voice -over, which is extremely hit or missed. . It is a classic speech with AI-generated language: Elements in a list are spoken in inconsistent tones. Some words are slow or artificially slowed down as if the bot had one or two drinks. But honestly, I had expected worse. This is not as if you are seeing a documentary that is told by the Tikok -ai. It’s not great (it’s not even good), but it offers one a little More emotions than I expected.
The topics extend to the AI dubs of the respondents of the documentary. In many cases, the voice is quite cumbersome, stilted and full of the inconsistencies mentioned above. Sometimes the voice is completely without emotions and sounds more like a voice -over tool from the past few years than after AI votes that we expected.
For Amazon’s Credit, it sounds as if the AI uses a different voice for every person who synchronizes it. This makes it a little easier to distinguish between the different speakers, and puts it off from hearing the same mediocre AI voice for the entire film.
Long
Long It is a little more difficult to judge: The film originally produced in English has a Ki -Dub in Spanish. Since I don’t speak Spanish, it is not so easy to select the robot quirks that your ear absorbs when you know the language.
Despite it, Long offered a different experience than The cidSince this is a film: you listen to how the AI synchronizes about actors. Like the Ki -Voice -Oover in the documentary, these AI actors are not as bad as I expected. I expected AI votes to deliver their lines completely flat, but sometimes there are some emotions. No robot gains an Oscar here, mind you: In fact, the voices can be quite cumbersome and missing, the script is too loud, abrupt or clear. But for some lines I am not sure whether I would assume that it is AI votes if I didn’t know it in advance.
Even in more intense scenes, the AI ”tries”. When the character cries, the AI increases its voice. It’s anything but perfect, but it’s an interesting experiment to see what the AI can do:
In one scene, two characters compete in a game of “Chubby Bunny”, in which every marshmallows have to add further into their mouths while they can still say “Chubby Bunny”. Sometimes the AI actually sounds steamed, and although it is not “realistic”, it is a nice gesture. (Perhaps human intervention will come into play here.) At other times, however, the AI sounds completely normal, while the mouths of the actors are clearly full of marshmallow.
Another interesting peculiarity: one of the characters speaks in French, so that the AI track cuts out and the original audio comes into play. It is tiring because you can hear the difference immediately – not just the different languages, but the difference in the way the actual audio of the film sounds compared to the sterile AI edition.
I would like to try this function again in a film produced in Spanish with an English AI dub. I can imagine that it would be easier for me to choose the problems and know whether more (or all) of the voices would sound completely wrong.
Is the future of synchronization of AI?
Look, there is no way that I would never enjoy seeing an entire film or a series with a Ki -Dub. Personally, I am not a big fan of dubs that are generated by people, so I have no interest in someone who sometimes sounds like a robot that shoots for the eGot.
But I have to admit that the technology is not terrible. In some places, the voices sound realistic enough for me to get the core, especially if the voice -over would also be monotonous and informative – like during a documentary. And it is interesting that these AI dubs can include elements from the show or in the film like subdued speech during a chubby rabbit game.
I don’t want a language actor who get out of work here. They deserve to be paid for their work and their audience, deserve a high -quality dub in the shows and films that they pay for the reputation. However, I think that KI synchronization could offer a compromise for the times when there is no DUB on the way: If a studio is not paying for a DUB anyway, it would be helpful to have a dub at all – even if it is on the robot side. This would certainly help to affect the audience that may not be able to use the subtitles.
Of course, it is naive to believe that studios will only use this technology for these specific uses instead of eliminating a whole human profession to save some money. For this purpose, I would be happy if the entire AI synchronization ends here and now: it is interesting, but it is not good, so do not replace real language players.
