Artificial intelligence in journalism: In newsrooms worldwide, there is a heated debate about what it means for the future of the media industry and its employees. First attempts to have texts written by AI-fueled systems are viewed with suspicion, and every mistake in the text is celebrated as proof that people are still indispensable in journalism.
At the same time, since they arrived in everyday journalistic life, automatic assistants have long and inconspicuously performed work that editors have always had to do, even though they have only a limited connection with journalism. This starts with spell-checking in word processing and editing systems, little helpers that already existed when artificial intelligence was first talked about, especially in utopian novels and films. Also used for years are transcription machines that translate spoken texts into written ones and translation programs that translate texts from one language into another. Conversely, text-to-speech applications pronounce what is available as text out loud: any smartphone can do this today, but announcements at airports or train stations are now predominantly generated automatically as well.
Transcriptions and translations have always been part of everyday life in multimedia editorial offices. plain X is a platform that combines selected AI applications into a package that makes sense for editorial offices and offers them under a uniform interface. With plain X, for example, it is possible to “listen to” the audio track of a video and compare it with the video’s manuscript, if necessary, automatically create subtitles and synchronize them with the audio track, translate the subtitles from one language to another and thus offer subtitles in two or more languages – or output the translation as a synthetic language and thus offer the video in another language.
Deutsche Welle is the manufacturing company Priberam’s development partner–which means extra challenges for the developers. For DW, the artificial intelligence must be available in all DW languages and support Arabic, Chinese, and several other fonts in addition to Latin and Cyrillic characters.
For almost a year, the author of this text has been working with plain X and how it copes with the requirements of Deutsche Welle. During this time, it was possible to observe how the different engines were evolving; External systems embedded in plain X, which at the beginning of the year still produced mediocre transcription or translation results, caught up with their more powerful competitors. Advanced features such as automatic speaker recognition have been retrofitted where available in the respective engines. The manufacturer was open to corresponding requirements and suggestions from Deutsche Welle; workflows that some users perceived as unnecessarily complex could be smoothed out, and the integration of plain X into DW’s system landscape with interfaces to the editorial planning and production system Open Media and the video editing system HIVE as well as to DW’s CMS also made considerable progress.