Who: Volkswagen Group & Azure AI
The Volkswagen Group serves customers worldwide and delivers a vast range of documents in over 60 languages. Each year, up to two billion words and documents need to be translated within the Volkswagen Group, and this volume is only rising. The advent of new communication streams and an increasing demand for faster translation turnaround times necessitated an enhancement and expansion of the existing translation systems. As a solution, Volkswagen turned to Azure AI, offering a faster, more precise, and cost-effective system with advanced learning capabilities.
What: High-Volume, High-Quality Text and Document Translation
Translating up to two billion words annually in more than 60 languages may sound like a formidable task, but it’s routine for the Volkswagen Language Services team. They are constantly immersed in translating a diverse range of texts and documents – from owner’s manuals to presentation decks to board of management documentation. Many of these documents, like the owners’ manuals, are legally binding, which underscores the need for the Volkswagen Group to ensure high-quality translation.
Before: Working hand in hand with the machine
For the past seven years, the Volkswagen Language Services has been recording a substantial uptick in translation volumes. This trend has only been accentuated by the Volkswagen Group’s push for digital transformation. Moreover, clients now expect results within shorter timeframes. However, given that the existing technology translates content in a rules-based manner, it requires investment each time a new language is added. This approach is not only expensive, but the rules-based technology also restricts the speed of translation and the total number of words that can be stored in the source and target languages.
The process: Building a custom, cloud-based platform
In order to address these challenges, Meyer-Aun’s team started looking for cloud-based alternatives that promised greater scalability and flexibility, courtesy of the neural machine translation models available in the cloud. After an exhaustive comparison of various providers, they selected Microsoft Azure.
After: Translating full documents and complex file types, fast
The first iteration of the translation system processed text translation, which has now been expanded to include document translation. As a result, the Volkswagen Group can quickly translate documents in multiple file types, such as .xls, .tlf, XML, PowerPoint, and others. The system currently translates an average of 325 documents per day, ranging from 1-page files to 1-million-character books.
Conclusion: Safeguarding compliance and confidentiality
A key deciding factor for the Language Services team was that none of the data—translation memories, documents to be translated, and trained models—was allowed to leave the European Union (EU) for data protection reasons. Ensuring compliance with the EU’s strict data protection regulations made Azure a natural choice. Looking ahead, the Volkswagen Group is keen on exploring custom translator capabilities and finding new ways to employ machine translation to meet the needs of customers and employees.
Leave a Reply
You must be logged in to post a comment.