RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

Sinterra Media: Services for localizing TV content using artificial intelligence

Product
The name of the base system (platform): Artificial intelligence (AI, Artificial intelligence, AI)
Developers: Sinterra Media
Date of the premiere of the system: 2024/07/16
Branches: Media, Television and Radio Broadcasting,  Telecommunications and Communications
Technology: Speech technology

The main articles are:

2024: Presentation of the service for localization of TV content

Sinterra Media has developed services for localizing TV content using artificial intelligence (AI). This will help broadcasters working in other countries to quickly translate libraries and comply with the necessary legislative requirements for localizing content. The company announced this on July 16, 2024.

There are rules for localizing TV content for foreign broadcasters in different countries. For example, in June 2024, the President of Kazakhstan signed the law "On Mass Media," according to which the mandatory volume of television and radio programs in the state language is established. Broadcasters are obliged to ensure the translation of at least 10% of television and radio programs of an educational, cultural, educational, sports, entertainment and leisure nature into the Kazakh language. This means that Russian broadcasters working in the republic need to localize their libraries as early as July 2024.

Sinterra Media has developed technological services using artificial intelligence to localize content: subtitling, translation in one-voice and polyphonic modes. The application of AI greatly simplifies this process. For example, if the editors of TV channels can take another month to translate the average monthly volume of content - which is about 600-800 hours - then using services developed by Synterra Media using artificial intelligence for several days. This reduces manual effort, minimizes human factors and improves overall productivity.

When subtitling during the Russian audio track at the bottom of the screen, Kazakh subtitles are launched, they can also be "sewn" into teletext. At the same time, AI correctly supports translation, preserving the meaning of sounding speech, takes into account the features of the movie video sequence, as well as local requirements for the image of subtitles on the screen.

With a one-voice translation (all characters speak with one voice), artificial intelligence allows you to maintain a high quality of translation and voice acting. {{quote 'According to the experts-linguists whom we invited to evaluate the service, the error level was only one error per hour. As a rule, these errors were associated with the reproduction of AI stresses. This is a very small level of error, - explained Oleg Fatun, Deputy General Director - Director for Business Development of Sinterra Media. }}

Also, when localizing content, you can use a polyphonic translation, when, for example, a woman speaks with a female voice, a man speaks with a male voice. With the participation of a group of characters in the frame at once, artificial intelligence is able to voice them with different voices and timbres. At the same time, in all variants of content localization, AI understands cultural nuances, idiomatic expressions and contextual references.

Sinterra Media is also testing another content localization service - natural translation into any languages ​ ​ of the world (lipsink technology, lip movement synchronization to the sound row) and tonality analysis. The decision will take into account the moment of correct assignment of lipsink to the actor. For example, so that the speech and movement of the lips of a person who does not speak in the frame are not superimposed on the actor who is in the frame at that moment.