Issue link: https://digital.copcomm.com/i/1505114
26 cgw j u ly • a u g u s t • s e p t e m b e r 2 0 2 3 sional translators create this translation data for media content and carefully review it through multiple QC steps. The outcome is a pure language learning material that we call "golden data." We have developed media-specific techniques, such as context awareness (CA), which enables our engines to understand and translate sentences with regard to the hidden context, providing a higher degree of accuracy in translating colloquial dialogues that re- quire capturing emotion and nuance. Furthermore, our engines employ multimodal techniques to con- sider additional forms of information to aid in the translation process. For example, translating a description of an image is more accurate when incorporating the image itself. The engine takes into account information relevant to each language. When there are multiple speakers in a video, factors such as age, gender, and relationships are also reflected in the translation to better convey the situation or relationship between individuals. C G W : H O W A R E X L 8 ' S T R A N S L A T I O N E N G I N E S T R A I N E D T H R O U G H D E E P L E A R N I N G T E C H N I Q U E S ? Tim Jung: As mentioned previously, one of the key ingredients of XL8's translation engines is the golden data. The domain, quality, and quantity of this data perfectly fit to train world-best machine translation engines specialized for media contents and colloquial dialogues. Based on this exclusive data, we develop our models to leverage state-of-the-art techniques in neural machine translation (NMT), large language models (LLM), and deep learning (DL). On top of that, we develop our own model architectures and training techniques specifically designed to support advanced features, in- cluding but not limited to, CA, formality-aware translation, personal- ized MT, and simultaneous translation. C G W : T E L L U S A B O U T T H E I M P O R T A N C E O F M A I N T A I N I N G C O N S I S T E N T T E R M I N O L O G Y ( C A T C H P H R A S E S , S L A N G , P L O T- S P E C I F I C R E F E R E N C E S E T C . ) W H E N T R A N S L A T I N G M E D I A C O N T E N T. H O W D O E S X L 8 M A K E T H I S P R O C E S S E A S I E R T O A C C O M P L I S H ? Tim Jung: Whether you choose to post-edit or not, using a glossary is another tool to save time and money by leveraging the power of AI to achieve the closest possible accuracy. It's a valuable tool that supports a successful viewing experience. This is why XL8 provides a glossary feature where customers may upload relevant terms and nomenclature specific to the content being transcribed or translat- ed. The use of glossaries is essential for two main reasons. Firstly, it helps improve accuracy in speech-to-text jobs by providing a refer- ence for spelling and word identification in the sync tool. This is es- pecially important in live events or scripted content where accurate language is crucial for a successful viewing experience. Secondly, glossaries are valuable in the translate tool as they al- low us to capture specific language pairs by genre, domain, or title. By offering multiple source and target pairings for each language,