Synterra Media, a digital partner in the production, processing, and delivery of media content, has developed services for the localization of TV content using artificial intelligence (AI). This will assist broadcasters operating in other countries to quickly translate their libraries and comply with necessary legal requirements for content localization.

Localization rules for TV content for foreign broadcasters vary by country. For instance, in June 2024, the President of Kazakhstan signed the "Mass Media" law, which mandates a required volume of TV and radio programs in the state language. Broadcasters are required to ensure that at least 10% of educational, cultural, enlightening, sports, entertainment, and leisure programs are translated into Kazakh. This means that Russian broadcasters operating in the republic must already localize their libraries.

Synterra Media has developed technologically advanced services using AI for content localization, including subtitling and translation in both single-voice and multi-voice modes. The application of AI significantly simplifies this process. For example, if editors at TV channels take about a month to translate an average monthly volume of content—approximately 600-800 hours—using the services developed by Synterra Media with the help of AI can reduce this time to just a few days. This minimizes manual efforts, reduces human error, and increases overall productivity.

During subtitling, Kazakh subtitles appear at the bottom of the screen while the Russian audio track plays, and they can also be embedded in teletext. The AI accurately maintains the meaning of the spoken dialogue, considers the visual elements of the film, and adheres to local requirements for subtitle display on the screen.

In single-voice translation (where all characters speak in one voice), the AI ensures high quality in translation and voiceover. «According to the assessments of linguistic experts we invited to evaluate the service, the error rate was just one mistake per hour, typically related to the AI's pronunciation. This is a very low error rate», explained Oleg Fatun, Deputy General Director and Business Development Director at Synterra Media.

Additionally, multi-voice translation can be used, where, for example, a female character speaks in a female voice and a male character in a male voice. When a group of characters is present on screen, the AI can voice them with different voices and tones. In all localization options, the AI understands cultural nuances, idiomatic expressions, and contextual references.

Synterra Media is also testing another content localization service—natural translation into any language worldwide (lip-sync technology, synchronizing lip movements with audio) and sentiment analysis. The solution will ensure correct lip-syncing for the actor. For instance, it will prevent the speech and lip movements of the off-screen speaker from overlapping with the actor currently on screen.