in All Things Localization

With the emergence of natural language generation (NLG) – a class of language AI that can not only read and translate, but actually create original content – video and entertainment localization are just one of many fields experiencing the disruptive impact of automation in areas long reserved for human talent. Yet, while there is a broad assumption that AI solutions will replace human linguists wholesale as the cheaper, more productive option, insights from the machine translation era of language services reflect a far more nuanced reality, in which technology is not so much a replacement as an enabler of services that can extend communications when and where they are needed. Entertainment currently faces a noted shortage of linguistic resources paired with studio production capabilities in relation to the vastness of global demand for localized films, TV shows, and game titles in world markets. When a small monopoly of localization studios reaches capacity, AI solutions are no longer a threat to existing production models, but a vital stopgap for overruns they cannot adapt to sustain. Simply put, technologies that can help everyday linguists deliver their local cultural understanding to screen are in high demand for multilingual video content. This week, Technology Review highlights the impressive impact of one NLG technology that in the localization world is likely to prove an unexpected boon from what could otherwise be a job-taker: digital voice actors.

As CSOFT noted in our recent white paper on streaming entertainment localization, 2022 brings focus to the role of language AI models in video dubbing and subtitling with the advent of startups using AI script transcription and dubbing to compensate for a voiceover talent shortage. But where automated dubbing inherently comes with the hazards of trusting machines with culturally nuanced material (requiring human linguistic review,) automatic source-language transcription and subtitling stand out as empowering tools for linguists that can scale their support capabilities to studio or near-studio efficiency. Similarly, digital voice actors are not necessarily a replacement for what an actual linguist would do to overdub a script, but rather a means to extend behind-the-scenes efforts from translators into direct focus without the additional intermediary of a studio facility. Seen one way, the most talented voiceover artist could now be someone physically incapable of speaking, with the potential for a translator to craft the message that an AI tool will bring to life. In short, it serves to diversity the potential talent pool of multimedia localization while reducing overhead, bringing new players to the competitive landscape of multilingual content creation.

Related:  Language Preservation, Climate Action, and LLMs: How AI Translation is Driving Positive Change

As Technology Review notes, ‘digital voice actor’ is somewhat of a euphemism for ‘deepfake’, and “…human voice actors aren’t going away just yet. Expressive, creative, and long-form projects are still best done by humans.” It is precisely the sheer milieu of multimedia content produced across languages, however, that makes the possibility of localizing it without human voice acting talent, without loss of quality, so appealing. Further, the potential to standardize the ‘character’ of a brand or campaign in a way that does not create a lasting dependence on individual humans is immensely valuable to companies leveraging multilingual marketing to compete on social media and other digital channels. For language service providers like CSOFT, language AI is a powerful tool that does not reduce the relevance of our linguists, but rather extends and optimizes it.

To learn more about CSOFT’s technology-driven translation solutions, contact us at!