Machine Translation

Increase quality and shorten the translation workflow


Transforming Machine Translation


Technology, as it has in almost every area of business, is playing a transformative role in machine translation. Recent developments in the use of Neural Networks to improve translation quality further highlight how far we have come. The introduction of a Neural Machine Translation (NMT) model, to challenge the widely-used Statistical Machine Translation (SMT) model, amounts to a sea change in the possible speed and quality of translation output, and unveils the unlimited potential that comes from using an artificial intelligence-based approach.

Regardless of the preferred method of machine translation, whether SMT or a future world driven by NMT, the need remains for a scalable, efficient way to manage the translation workflow. There is still no replacement for subject matter experts who manage the post-edit and final review processes. These reviewers ensure that the nuances of the language are properly reflected, and the final message is in context with the cultural norms and values of regional audiences.

The quality of the resulting translation system gets closer to that of average human translators
Google Translate Group
The current approach of statistical, phrase-based MT has kind of reached the end of its natural life
Alan Packer
Head of the Language Technology, Facebook

Machine Translation Deployment


The Cloudwords platform integrates with a range of machine translation engines, including Lilt, Google Translate and MSFT Translate. We support several deployment and workflow options that meet the translation requirements of our customers. These include: using machine translation alone, or combining machine translation with a post-editing process that can either integrate with a Language Service Provider (LSP) or combine the human review efforts of an LSP with an internal subject matter expert. Further workflow benefits can be gained by leveraging Cloudwords Translation Memory to reduce time to market, verify that the same words aren’t re-translated, and ensure the final translation is on-brand and adheres to company specific nomenclature. 

Cloudwords was founded to help leading enterprises drive global growth and competitiveness through a highly efficient localization process. We will continue to provide best-in-class integrations into marketing technologies, and leverage leading translation methodologies to enable our customers to take full advantage of any new developments in the localization industry. 


Cloudwords Machine Translation Workflow

  1. Workflow begins with an existing marketing automation system, Web CMS, file sharing platform or knowlegebase

  2. Content from the source system is automatically extracted by Cloudwords and is machine translated, leveraging the customer's Translation Memory

  3. The field marketing reviewer, post-editor or subject matter expert reviews the content to be sure it is on brand, on message and properly localized

  4. Once the content is reviewed and post-edits are completed, content is labeled as approved and machine translation updates the Translation Memory

  5. Cloudwords automatically publishes everything back to the source system and content is now ready to be posted, distributed or executed as a campaign