in ,

Using AI and machine learning to solve the challenge of globalization in the entertainment industry!

According to recent studies, Facebook barely detects 3% to 5% of dangerous information on its network.

Using AI and machine learning to solve the challenge of globalization in the entertainment industry the current debate over mistranslation in the Netflix smash “Squid Game” and other films underscores the difficulties that technology faces when publishing material that spans languages and cultures throughout the world.

Every year, tens of thousands of movies and TV episodes are created in the worldwide media and entertainment sector, with hundreds of streaming platforms displaying them in the hopes of reaching a global audience of 7.2 billion people living in roughly 200 countries. No one in the crowd speaks any of the nearly 7,000 languages that are recognised. Subtitles and audio dubs must be created for worldwide distribution if the intention is to release the video abroad.

Creating “subs and dubs,” also known as “localization,” has been a human-centered process for decades, in which someone with a thorough understanding of another language sits in a room, reads a transcript of the screen dialogue, watches the original language content (if available), and translates it into an audio dub script. It’s fairly uncommon for this process to take several weeks from start to completion, depending on the language.

The screenplay is then performed by voice actors who make every attempt to replicate the motion and lip movements as precisely as possible once the translations are completed. Following the final cut speech, audio dubs are created, and then subtitles are made from each audio dub. Any concessions made in the language translation may be vulnerable to further concessions in the subtitle development. It’s simple to understand how a tale may be mistranslated or changed.

Because certain words, acts, or settings are not uniformly translatable, the most diligent localization process includes some amount of cultural understanding. Before starting work, Bong Joon-ho, the director of the 2019 Oscar-winning picture “Parasite,” issued specific remarks to his translation team. Limitations of time, available screen space for subtitles, and the requirement for cultural awareness, according to Bong and others, further complicate the process. Nonetheless, when done effectively, they lead to greater degrees of cinematic enjoyment.

Because of the exponential rise of distribution channels and the constant influx of new material, people involved in the localization process are looking for innovative techniques to speed up production and improve translation accuracy. Artificial intelligence (AI) and machine learning (ML) are popular solutions to this challenge, but none has yet proven to be capable of completely replacing human localization. Directors of films like “Squid Game” and “Parasite” have yet to take that step. This is why.

Culture is important.

For starters, a literal translation is incapable of capturing every linguistic, cultural, or contextual detail in the script, accent, or action. Machine-based translations are typically referred to as “more like dictionaries than translators” by AI businesses, who also remind us that computers can only do what we teach them while claiming that they lack comprehension.

“Red Light, Green Light,” for example, is the English title of the first episode of “Squid Game.” This is the title of the children’s game that was shown in the first episode. The original Korean title is “무궁화 꽃이 피던 날” (“Mugunghwa Kkoch-I Pideon Nal”), which directly translates as “The Day the Mugunghwa Bloomed,” which has nothing to do with the game they’re playing.

The title represents new beginnings in Korean culture, which is the game’s heroes’ pledge to the winner. Although “Red Light, Green Light” is connected to the episode, it omits the larger cultural allusion of a promised fresh start for those who have fallen on hard times, which is a major topic in the series. Some may argue that calling the episode after the game played isn’t a significant concern because the cultural metaphor of the original title is lost on the translators, but it is.

How can we expect robots to perceive and use these distinctions independently if people do not make the link and do so themselves?

Learning vs. knowledge

It’s one thing for a machine to translate Korean into English, but it’s quite another for a human to do so. It’s another thing entirely to understand the distinctions in connections in “Squid Game,” such as those between immigrants and natives, strangers and family members, employees and bosses, and how those interactions affect the tale. It’s difficult enough to programme cultural knowledge and emotional awareness into AI, especially when such feelings are expressed without words, such as a look on someone’s face. Even so, it’s difficult to predict emotional facial responses that vary by culture.

When it comes to explainability, interpretability, and algorithmic bias, AI is still a work in progress. Given where the industry stands in terms of implementing AI/ML, the concept that robots would self-train is a stretch. Context is everything in a content-heavy, creative sector like media and entertainment; there’s the content creator’s presentation of context and then there’s the audience’s interpretation of it.

Furthermore, context equals culture in terms of worldwide dispersion. A system achieves digital nirvana when it can orchestrate and forecast audio, video, and text, as well as the various levels of cultural nuance present at every given frame, scene, theme, or genre level. It all starts with high-quality training data — essentially, a data-centric approach rather than a model-centric approach.

According to recent studies, Facebook barely detects 3% to 5% of dangerous information on its network. Programming AI to grasp context and purpose, even with millions of dollars available for development, is extremely difficult. While fully autonomous translation solutions are still a ways off, AI and machine learning can help reduce workload now. It’s possible.

A two-step human and AI/ML method can give the deep insights needed to identify content that any country or culture may find unacceptable by analysing millions of films and TV series paired with the cultural expertise of individuals from almost 200 countries. This cultural roadmap is then utilised in the localization process to maintain plot consistency, minimise cultural blunders, and secure worldwide age ratings, all while reducing post-production time and costs without putting the project at danger of regulatory trouble.

Today’s audiences have more content options than ever before. To succeed in the global market, content providers must pay greater attention to their audience, not just at home but also in other countries.

Working with firms that understand local audiences and what important to them is the quickest way to success for content creators and streaming platforms, ensuring that their material is not lost in translation.

Written by IOI

Get the latest stories from Tech & Innovation from around the globe. Subscribe Now!

Leave a Reply

Your email address will not be published. Required fields are marked *


Is Facebook working on a new AI model that predicts future actions?


The best free AI and data science courses that were launched in 2021!