[Digital Today reporter Seulgi Son] "Korean content is already competitive enough. The problem is that overseas viewers drop off because of language even after they discover the content."

Shin Hyun-jin (신현진), CEO of HudsonAI, said in a recent interview with a reporter that he sees language and cultural barriers as a major bottleneck in exporting K-content and stressed that he aims to solve the issue with AI. Even with strong content, it is hard to translate discovery into actual viewing if local audiences do not read subtitles or prefer dubbing in their own language, he explained.

HudsonAI is a company that uses AI-based translation and speech synthesis to localise video content into more than 80 languages. It is expanding its applications to include sports and kids content as well as films and dramas.

Shin pointed to South America and Western Europe as markets with high demand for dubbing. He said South Korea is accustomed to watching with subtitles, but many overseas viewers prefer audio in their native language. Dubbing is effectively essential for kids content, whose main audience is children.

He said, "In South Korea, people consume content even if only subtitles are provided, but overseas, whether it is dubbed has a big impact on viewing performance," adding, "There are many cases where companies have good content but cannot expand because of localisation costs."

The expansion of the ad-supported free streaming TV (FAST) market is also increasing demand for AI dubbing. FAST needs to keep supplying large volumes of content to multiple countries, but under conventional methods, translation, voice actor recording and quality inspection costs rise together each time the number of languages increases.

Shin said, "FAST earns less revenue per title than subscription-based online video services (OTT), so it is hard to bear the existing dubbing cost structure," adding, "To supply multiple countries, localisation costs must be lowered with AI."

HudsonAI uses AI to distinguish speakers in video, translate lines and then generate voices that match roles. Shin said this can cut existing dubbing production costs to less than half at minimum and also shorten production time.

Still, AI cannot fully replace the entire dubbing process. Human review is needed for a character's emotions, the context of scenes and expressions that fit local culture.

HudsonAI is developing "agent localisation" to maintain quality while raising the level of automation in the localisation process. Multiple AI agents responsible for translation, voice generation and quality inspection exchange outputs to check and correct errors. AI handles repetitive work, while experts verify final quality and cultural context.

Demand for AI localisation is expanding beyond films and dramas to corporate product introductions, training and promotional videos. HudsonAI is also offering a localisation solution that can convert videos owned by companies into multilingual subtitles and audio.

Shin said, "AI application methods must also differ depending on content genre and purpose of use," adding, "Demand is growing to reduce cost and time not only in FAST, sports and kids content but also in corporate videos."

Verifying overseas distribution system for AI dubbing with K-FAST

HudsonAI is participating as a lead technology company in the "Global K-FAST Alliance" promoted by the Ministry of Science and ICT. It is tasked with dubbing domestic content to fit language and culture by country so it can be supplied to overseas FAST channels, and with continuously securing localised content.

The goal is to go beyond translating one or two individual works and build a system to repeatedly produce large-scale content in multiple languages needed to operate FAST channels. Because FAST channels are divided by country and must maintain 24-hour programming, sufficient localised content must be secured for each target region.

Shin said, "FAST must continuously supply content to country-specific channels, so localisation speed and cost determine business viability," adding, "Through this project, it is important to show whether AI dubbing can lead to actual overseas viewing and distribution results."

He also said project performance should not be assessed only by the number of channels launched. Even if technology companies handle localisation, the final outcome can vary depending on platform programming and marketing and local distribution capabilities.

Shin said, "Performance should be evaluated not by how many channels are created but by actual viewing volume and content distribution results," adding, "Government support should go beyond technology development and move toward creating initial demand so content companies and platforms actually purchase and use AI localisation services."