Japanese Manga Translation Startup Orange Raises USD 19.5M

Japanese Manga Translation Startup Orange Raises USD 19.5M

Japanese manga localization tech company Orange raised JPY 2.2bn (USD 19.5m) in what they called a “pre-Series A” financing, the company announced in a May 6, 2024 press release.

Manga publisher Shogakukan led the financing round. Other participants included the Japanese government’s Japan Industrial Innovation Investment Corporation, Globis Capital Partners, ANRI, SBI Investment, JIC Venture Growth Investments, Miyako Capital, Chiba Dojo Fund, Mizuho Capital, Mitsubishi UFJ Capital, and GFR Fund.

Orange, founded in 2021, uses computer vision and NLP, including AI translation, to translate manga titles in a fraction of the time required by traditional methods.

The company claims that it can translate, from Japanese to English, up to 500 manga titles per month — a figure Orange pegs at about five times the current industry rate. 

The company’s ultimate goal: translate 50,000 manga titles in the next five years. Orange will use some of the funds raised in this round to launch a digital manga store called emaqi in summer 2024. The store will reportedly feature recommendations from manga influencers — as well as AI-generated suggestions.

Orange stated in its press release that only approximately 2% of manga released annually in Japan has been officially translated into English, “partly due to the difficult and lengthy translation process and the limited number of translators.” 

The lag time between Japanese-language releases and their localized counterparts is a contributing factor to the global piracy market, which the Content Overseas Distribution Association estimated at USD 5.5bn in 2022. CODA and other anti-piracy groups believe AI translation could reduce that lag time and, in turn, the demand for unauthorized localized versions. 

Speed vs. Quality vs. Price

Katrina Leonoudakis, an experienced Japanese-English translator and localization producer at Deluxe Media, doubts that this strategy will pan out, “because foreign language readers will not pay money for bad translations.”

Leonoudakis explained that Japanese poses a particular challenge for AI because it is a high-context language, meaning that it often omits important contextual clues from text. Readers or listeners must infer information — such as who is speaking and which plot points are being foreshadowed — based on the surrounding context. 

“AI isn’t smart enough to understand context like that just yet, so its output will be extremely messy and full of mistakes,” Leonoudakis told Slator. “Other companies in Japan have tried to use machine translation to quickly publish translated manga, but the output has been of extremely low quality. The translation is not only riddled with errors, but it doesn’t sound like good English, either.”

Orange subtly acknowledges the limitations of AI translation, since its method does require a human translator to correct the AI-generated output. 

But according to industry news website CBR, translators have claimed to have been fired and then rehired “under worse terms so as to work with AI […] leading to professionals doing the same work for less compensation.” (CBR did not state whether this applied specifically to translators working with Orange.) 

For the time being, Orange plans to translate titles from Japanese to English for both children and adults, and may consider expanding into India and Spanish-speaking markets in the future.